Cracks In Concrete Structures (CICS) Dataset

Published: 16 June 2023| Version 1 | DOI: 10.17632/9brnm3c39k.1
Contributors:
,
, Mahdi Shojaei

Description

A drone (phantom4 pro) with a 20 M Pixel and full HD camera was used to prepare images of concrete structures in campus of Imam Khomeini International University and Qazvin city in Iran. Various of concrete structures, using concrete cement type 2 and grade 250 to 300 were considered to collect images. The information provided by phantom4 pro was saved as color images and videos which resulted in the storage of 900 images with 4k resolution. The stored color images were first converted to grayscale images, and then, each of the 900 grayscale images was divided into small images with 330×330 pixels resolution. Then, 12000 grayscale images with the size of 330×330 pixels were created. These images were classified visually into three classes (without crack, simple crack, and multibranched crack). Finally, these images named as Cracks In Concrete Structures (CICS) dataset consists of three folders. The folder names reflect the shape of the contained images. There are 4000 images without crack, 4000 simple crack images, and 4000 images with multibranched crack.

Files

Institutions

Imam Khomeini International University

Categories

Machine Learning, Image Classification, Crack, Concrete Cracking, Deep Learning

Funding

Imam Khomeini International University

11805

Licence