Classification of Cracked Concrete Images Using Convolutional Neural Algorithm

Abstract

Concrete is the most widely used construction material in the construction industry. This concrete material is one of the important materials used in infrastructure development in Indonesia. Cracks in a building is something that is avoided in every infrastructure development. Cracking is a condition when water evaporation in concrete occurs very quickly due to changes in weather. Many traditional measures have been taken to determine if a building is cracked or not cracked, but it has resulted in inaccurate conclusions. Indonesia is a country that is doing a lot of significant infrastructure development to become a developed country and improve the country's internal situation. Therefore, one of the basic things in a construction that uses concrete in infrastructure development is very concerned about its condition. Based on these problems, this study uses one method of classifying an object using digital images, namely Convolutional Neural Network to classify cracks in concrete. This study aims to classify images of concrete cracks using the Convolutional Neural Network algorithm. The stages in this study are the preparation of secondary data research data collection which is stored in a file format (.jpg) with tens of thousands of data and the second stage is the pre-processing stage by selecting a class on the image, then compiling a design using the Convolutional Neural Network architecture, training and testing process will be carried out from the results obtained, and accuracy testing will be carried out. The results of this study are expected to provide benefits from seeing the results of the best level of accuracy of classifying concrete crack images using the Convolutional Neural Network algorithm.