Enhanced Concrete Surface Crack Detection using Deep Learning

• Author(s): Harsh Gupta ; Namita Goyal ; Vandana Choudhary
• Paper ID: 1704685
• Page: 566-571
• Published Date: 18-06-2023
• Published In: Iconic Research And Engineering Journals
• Publisher: IRE Journals
• e-ISSN: 2456-8880
• Volume/Issue: Volume 6 Issue 12 June-2023

Abstract

This research paper presents a novel approach to detect cracks in concrete surfaces using the ResNet50 deep learning model with advanced image transforms. The proposed methodology demonstrates high accuracy in detecting cracks compared to traditional methods. In this study, we have employed transfer learning to train the ResNet50 model on a large dataset of concrete surface images with cracks. Additionally, we have applied advanced image transforms, including random rotation, random brightness adjustment, and random scaling, to enhance the accuracy of the model further. The results of this research provide a promising solution for automating the process of crack detection in concrete surfaces, which is a critical step in ensuring the structural integrity of infrastructure. The dataset is segregated into two distinct classes: negative and positive, which correspond to images of intact concrete surfaces and concrete surfaces with cracks, respectively. Each class consists of 20,000 images, resulting in a total of 40,000 images. The images are in RGB format, with a resolution of 227 x 227 pixels. These specifications provide researchers with a balanced dataset to use for image classification tasks aimed at detecting cracks in concrete surfaces.

Keywords

Concrete surface crack detection, crack detection, deep learning, Image Processing, Convolutional Neural Network, CNN, Resnet 50

Citations

IRE Journals:
Harsh Gupta , Namita Goyal , Vandana Choudhary "Enhanced Concrete Surface Crack Detection using Deep Learning" Iconic Research And Engineering Journals Volume 6 Issue 12 2023 Page 566-571

IEEE:
Harsh Gupta , Namita Goyal , Vandana Choudhary "Enhanced Concrete Surface Crack Detection using Deep Learning" Iconic Research And Engineering Journals, 6(12)