Crack on Brick Wall Detection by Computer Vision using Machine Learning

Age and human activity are eroding old masonry constructions. Historic buildings need regular examination and care to maintain their integrity; Visual inspection is expensive, time-consuming, and subjective. This paper proposes an autonomous image-based fracture identification method for masonry constructions. Previous crack detection methods used hand-crafted characteristics categorized by algorithms. This method mainly depends on vectors and it may fail due to extract hidden structures. This research integrates deep CNNs and SVMs to identify cracks (SVM). CNN extracts feature from RGB pictures, and SVM replaces a SoftMax classifier to improve the classification. Photographed Flaws in the masonry were using a visual aids and drones at ancient locations. The photos trained and validated the system. The combined CNN and SVM model detects objects in validation photos with an accuracy of 86%. The technology can automatically identify flaws in masonry photographs, which is essential for inspecting historic sites.