FLOOD-DEPTH-ML: Machine learning-driven python application for estimation of urban flood depths through submerged vehicles detection

Climate change has caused an increase in floods worldwide that affect the lives of people and cause extensive property damage in urban areas due to high flood depth levels. Machine learning-based computer vision applications have been extensively used for the estimation of flood depth levels in urban environments. However, most applications are restricted to research purposes with their on-site usability remaining low and often fail to communicate the flood risk to the public in simple terms. In this study, we present a Python application that uses backend a you look only once (YOLO)-based detector with a simple graphical user interface (GUI) to classify vehicle inundation levels in five classes and help communicating flood risk. The Python application called FLOOD-DEPTH-ML available as open access allows users to analyze flood images/videos, online YouTube links, and most importantly its webcam feature, which users can use to easily integrate it with monitoring cameras to provide early warning for flood depths based on car submergence levels.