Tiny Machine Learning Models for Urban Flood Detection on Edge Devices

• Grant number: 25/20968-7
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: January 01, 2026
• End date: December 31, 2026
• Field of knowledge: Physical Sciences and Mathematics - Computer Science - Computer Systems
• Principal Investigator: Caetano Mazzoni Ranieri
• Grantee: Guilherme Bisse Alves Rodrigues
• Host Institution: Instituto de Geociências e Ciências Exatas (IGCE). Universidade Estadual Paulista (UNESP). Campus de Rio Claro. Rio Claro , SP, Brazil

Abstract

The increasing frequency and intensity of hydrological disasters in urban areas demands the development of monitoring and alert systems. In this context, research has explored different methods for early flood detection, including the use of Internet of Things (IoT) sensors to create intelligent monitoring systems. This project aims to develop and validate a flood detection system based on images, designed to operate even under unstable connectivity conditions, by leveraging the potential of Tiny Machine Learning (TinyML). The proposal focuses on optimizing computer vision models to be compact and efficient enough to run on autonomous embedded devices, such as the ESP32 microcontroller, enabling data processing directly at the network edge. For this purpose, a previously collected hydrological image dataset will be used, allowing for the training and evaluation of models adapted to hardware constraints. The goal is to enable local and autonomous inference, reducing reliance on cloud communication and increasing system resilience in critical situations. (AU)