Diseño de un prototipo de sistema de alerta temprana para inundación y desbordamiento, usando visión artificial y redes celulares para la ciudad de Puyo.

Abstract

A prototype of an early warning system for flood and overflow was designed using artificial vision and cellular networks for the Puyo city, to inform about fluvial activity, simulating the Citayacu River situations located in the Obrero neighborhood in the Puyo city. The system based its processing on a small board computer, Raspberry Pi3B +, together with a PiCamera module, which was positioned in such a way that images of the environment to be analyzed are captured. These images provide height information in relation to the levels maintained during normal fluvial activity. The images are processed by encoded segmentation algorithms under Python programming language in conjunction with OpenCV artificial vision libraries. In order to use processing techniques to locate in the capture, the reference plate, subsequently the contours corresponding to the strips related to identified levels, to report on the situation that was generated. Based on these data, information e-mails are sent to emergency groups, to users who are in flood risk zones, near the monitoring site. All the information collected is sent to a web server, allowing informing those who wish to know the situation that was presented in relation to the fluvial activity of the location. When the emergency groups decide to communicate certain information, it is sent to the device and retransmitted to different users, as well as published on the website. The device presented an alternative monitoring system for conventional monitoring procedures carried out by risk groups in the locality, in addition to continuous information on the web. The prototype is projected to evolve towards new systems that allow us to acquire data about other situations of interest for monitoring, giving us more information about possible natural disasters in each monitored location.