Water flow Detection Using Deep Convolutional Encoder-decoder Architecture

Due to shortcomings in sewage systems, there are areas in many cities that are highly susceptible to flooding in heavy rain scenarios. While the CCTV cameras for variety reasons scattered everywhere around the city, detecting sudden developed situations still require human attention and focus. Especially in emergency disaster situations, necessary actions that need to be done should also he quick in order to minimize the impact. Detecting flooding early on can save human lives, time, money for the government, as well as an important step to move towards smarter cities. In this paper, the authors propose a deep learning approach to detect floods in certain susceptible areas due to numerous reasons. Using state-of-the-art architecture capable of image segmentation, the system is trained to detect floods using either CCTV or PTZ-capable cameras, and the alert center is warned about the flooded area in order to divert units to respond accordingly to the situation.