Concrete Crack Images for Classification
Özgenel, Çağlar Fırat

@article{ozgenel2019concrete,
title= {Concrete Crack Images for Classification},
doi= {10.22260/ISARC2018/0094},
journal= {Mendeley Data},
author= {Özgenel, Çağlar Fırat},
year= {2019},
url= {https://data.mendeley.com/datasets/5y9wdsg2zt/2},
abstract= {Images of concrete surfaces from buildings on the METU campus, split into cracked and uncracked classes for crack classification.

## Contents

* 40,000 images, 20,000 per class (positive and negative)
* 227 x 227 pixel RGB images
* Generated from 458 high resolution images (4032 x 3024 pixels)
* Varied surface finish and lighting, no augmentation applied

## Related papers

* Özgenel, Ç. F. and Gönenç Sorguç, A. *Performance Comparison of Pretrained Convolutional Neural Networks on Crack Detection in Buildings.* ISARC 2018, Berlin. [https://doi.org/10.22260/ISARC2018/0094](https://doi.org/10.22260/ISARC2018/0094)
* Zhang, L. et al. *Road Crack Detection Using Deep Convolutional Neural Network.* IEEE ICIP 2016. [https://doi.org/10.1109/ICIP.2016.7533052](https://doi.org/10.1109/ICIP.2016.7533052) (the image generation method follows this paper)

## Source

Mendeley Data, DOI 10.17632/5y9wdsg2zt.2: [https://data.mendeley.com/datasets/5y9wdsg2zt/2](https://data.mendeley.com/datasets/5y9wdsg2zt/2). License: CC BY 4.0.

**Reference:** Özgenel and Sorguç, Performance Comparison of Pretrained Convolutional Neural Networks on Crack Detection in Buildings. 2018. https://doi.org/10.22260/ISARC2018/0094},
keywords= {Image classification, concrete, crack detection, civil engineering, structural health monitoring},
terms= {},
license= {https://creativecommons.org/licenses/by/4.0/},
superseded= {}
}