CrackVision: A Fusion of Deep Learning and Image Processing for Comprehensive Wall Crack Detection and Analysis

• Published in: 2024 3rd International Conference on Artificial Intelligence For Internet of Things (AIIoT) pp. 1 - 5
• Main Authors: S, Yaythish Kannaa G, D, Dennis Joshua, Chellamani, Ganesh Kumar, K, Manoj Prabhakaran
• Format: Conference Proceeding
• Language: English
• Published: IEEE 03.05.2024
• Subjects: Accuracy; Buildings; Computational modeling; Deep learning; Edge Detection; Image edge detection; Image processing; Jetson Nano; Neural Networks; Object Detection; YOLO
• DOI: 10.1109/AIIoT58432.2024.10574570

Wall crack detection is an essential part of the maintenance of building and safety as it allows for identifying and assessing defects on the wall surface. This information is necessary for determining the extent of the crack and the correct repair or reinforcement measures to be taken. Early methods, such as visual inspection, have been used, but recent advances in technology, such as infrared thermography and acoustic emission testing, have proven the detection of the crack. But in recent days, machine learning and computer vision techniques have become potent tools for automating and enhancing wall crack detection processes.This paper presents wall crack detection, combining image processing and deep learning. The edge detection algorithms like canny edge detector and You Only Look Once (YOLO) object detection models are implemented on the high-performance Nvidia Jetson Nano platform. This combination achieved an impressive 97.3% accuracy, demonstrating the effectiveness of the implemented methodology. Not only does the YOLO model increase the efficiency of wall crack detection, but it also improves the overall reliability of inspection procedures. This research is a step forward in utilizing cutting-edge technologies to solve critical building maintenance and safety aspects of the building.