Gathered and generated database for crack detection on concrete and brick walls

Published: 22 May 2023| Version 1 | DOI: 10.17632/vr37ss8xm6.1


Database was used for Master's thesis work "Identification of Asset Integrity Issues using Image Based Data Analysis and Deep Learning Methods" by Andrius Ambrutis and supervised by Mayur Pal. Data contains pipeline model for cracks imitations, two classes of images (cracks and not cracks) as well as code and Matlbab .mat file with trained network on 40% of entire data. Part of data in this dataset was generated using data gathered from: 1. ÖZGENEL, Ç.F. and GÖNENÇ SORGUÇ, A. Performance Comparison of Pretrained Convolutional Neural Networks on Crack Detection in Buildings. Surface Crack Detection. ISARC 2018, Berlin. Online. 2018. [viewed 2023-04-30]. Available from: <>. 2. Pexels. Brick Wall Cracks Photos, Download The BEST Free Brick Wall Cracks Stock Photos & HD Images. Online. [viewed 2023-04-30]. Available from: <>. In case of citation these sources should be mentioned as well (1 and 2) also usesage of these images should note violate lisences of earlier memtioned sources. In addition to dataset, we include 3d pipeline model created in Blender which can be used to generate data for almost any environment.


Steps to reproduce

All needed information and codes for model training are in thesis "Identification of Asset Integrity Issues using Image Based Data Analysis and Deep Learning Methods", which can be accessed from KTU database. For more additional information feel free to contact Andrius Ambrutis or Mayur Pal.


Kauno Technologijos Universitetas


Machine Learning, Cracking, Concrete Cracking