CRACKS DETECTION IN WALLS USING MACHINE LEARNING TECHNIQUES

Abstract

This paper presents a method for detecting cracks in walls using machine learning and deep learning techniques. Here are two types of image datasets that contain cracked images and non-cracked images. The dataset is named 'positive' and 'negative'. The positive dataset contains images with some cracks, and the negative dataset has images without cracks. Machine learning techniques are used for detecting cracks in the images. Machine learning encompasses three primary learning techniques: unsupervised learning, reinforcement learning, and supervised learning. In supervised learning, the user trains a program on known and labeled data, while in unsupervised learning, the user trains a program on unknown and unlabeled data. In this paper, we have examined unknown and unlabeled data of wall images. We have trained a program using a convolutional neural network to identify patterns within the data. A CNN is used to learn patterns within images, determining whether an image contains a crack or not. A CNN is a class of artificial network that plays an important role in various computer vision tasks. CNN uses the backpropagation concept by applying multiple building blocks, such as convolutional layers, Rectified Linear Units (ReLU), pooling layers, and fully connected layers.