File size: 528 Bytes
7b906db | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 | ---
configs:
- config_name: default
data_files:
- split: train
path: train/**
- split: test
path: test/**
---
# Defect VLM Dataset
This dataset contains paired structural defect images and English natural-language descriptions.
Fields:
- `image`: the defect image
- `text`: the generated English defect description
Splits:
- train: 884 images
- test: 211 images
The descriptions were generated from defect category and extent annotations while preserving the original train/test split.
|