This represents a prototype application designed to recognize cracks in glass and concrete RGB images using machine learning. Here's an overview of its key steps: 1- Initially, individual frames or images extracted from a video undergo preprocessing, during which edge segments are isolated using the Edge Drawing (ED) algorithm. 2- Following this, curvature features are derived from these edge segments, and a curvature histogram is generated. 3- A Support Vector Machine (SVM) classifier is trained using this histogram to distinguish between images with cracks and those without cracks. 4- Subsequently, the trained model is used to predict the class (cracked or non-cracked) of any new input images.
This prototype is implemented using C#.NET, built on WinForms, and developed in Visual Studio 2022. You can clone the repository onto your local computer and start the project using Visual Studio.
