Computer Vision Algorithm for Identifying Rivers as Indicators of Potential Flood (#781)

Authors

Rivera Funez, Christian Edgardo

Carrasco Bardales, Alberto Max

Abstract

The project proposes the development of a computer vision algorithm using YOLO (You Only Look Once) and the Roboflow training platform to identify rivers and predict potential overflows. This solution aims to address the issue of flooding in Honduras, where flood seasons are a constant threat, but the ability to take preventive measures is limited. The algorithm will be trained with images and river data to achieve accurate detection of bodies of water. Subsequently, it will be enhanced to identify early signs of overflows, enabling timely alerts to be issued. Early detection of overflows will allow for preventive measures to be taken, safeguarding lives and property. This computer vision approach holds the promise of providing an effective solution to anticipate and manage the risk of flooding in Honduras, where prevention and early warning are crucial to mitigate the devastating effects of these natural events. Keywords- Flood Prediction; CNN; Algorithm; Overflow.