AI Based Crack Detection and Correction System

Civil infrastructures need proper intervention to guarantee their durability and their service life. The damages in these structures such as cracks and crevices are caused by several triggers such as earthquakes, temperature variations, corrosion, rainfall etc. These damages appear on the surfaces of the structures. Traditionally, to assess the cracks, consulting companies sent their highly experienced engineers to go to the site and try to localize the damage based on naked eyes which was time consuming and even some areas were hampered by demanding access. In the last two decades there has been a use of impact hammering and ultrasonic equipment which are non-destructive evaluation methods. These tools are local and need prior information of the vicinity of the damage so that the area of the crack can be recognized exactly. The aim of this project is to develop a system that detects the crack in the early stage. This paper uses image processing and Convolution Neural Network (CNN) techniques to detect the crack. Further, several parameters like length, width and depth of the crack are calculated. This model also provides a repair solution for the crack detected. The proposed method of detection and analysis of the cracks on concrete walls is very useful especially in metropolitan cities where there are a large number of multistoried buildings, in the unreachable regions of the building, exteriors, in the interiors of the chimney etc. The cameras mounted on the buildings at various locations are used to capture photos with ease, and image stitching takes place. The captured data is given as an input to the model for detection. If the crack is detected, image processing operations are performed. Crack detection is vital for any structural health monitoring because, if these cracks are neglected, then it might lead to weakening of the structure and have catastrophic effects.