TRIB Crack Dataset

Published: 22 October 2024| Version 1 | DOI: 10.17632/gd5t95rv5t.1
Contributor:
Dumitru abrudan

Description

TRIB crack dataset was generated by the authors this study and was used a professional digital camera. During the capturing process, 7 streets from the capital of Romania, respectively Bucharest were inspected at different time of the day (morning, noon and evening). To overcome all the needs required in the process of deep learning methods the captured images contain different types of road cracks such as longitudinal, transversal, block, alligator and a various type of artefacts suck as oil stains, road markings on the asphalt, leaf and so on. The images were taken from 100 centimeters above the road surface and the dataset consists of 137 RGB images. To make the images suitable to be used in a deep learning methods, the images were divided into small images with a size of 256x256 pixels, being also applied different method of image augmentation

Files

Categories

Crack

Licence