By using multi-baseline stereo vision system to capture the photos of concrete structures periodically, this project aims to built a model that is able to detect cracks on concrete structures remotely and automatically using OpenCV, which will then extract relevent information such as width, depth, and precisely locate the cracks postion.

Certain machine learning methodologies had been applied in order to predict the cracks growth direction, so that effective measurements and mechanism can be taken when a certain treshold is triggered and thus securing the safety of the structure.