Video Surveillance-Based Urban Flood Monitoring System Using a Convolutional Neural Network

The high prevalence of urban flooding in the world is increasing rapidly with rise extreme weather events. Consequently, this research uses an Automatic Flood Monitoring System (ARMS) through a video surveillance camera. Initially, videos are collected from camera and converted into frames. After converting frames, water level can be identified by using Histogram oriented Gradient (HoG), which used to remove functionality. Completing extracted features, frames enhanced median filter unwanted noise image. next step classifiers Convolutional Neural Network (CNN), utilized classify images. performance analysis method analyzed various parameters. accuracy proposed 11% higher than that k-Nearest Neighbors (KNN) 5% ANN classifiers, processing time 7% less KNN 4% Artificial (ANN) classifiers.