Poster: Human Presence Detection After Earthquakes: An AI-Based Implicit User Interface on the Smartphone

Enrico Bassetti, Gianmarco Cavallaccio, Maria De Marsico, Emanuele Panizzi

Abstract: One of the main challenges of rescue operations after a devastating earthquake is the timely location of people trapped under debris. We propose a system that exploits the smartphone to detect the presence and implicitly interact with a person trapped in buildings. It leverages the phone microphone to detect sound waves generated by human breathing, heartbeat, and movement. It analyzes the signals on the smartphone itself using deep learning. A server collecting the results can support search-and-rescue operations or trigger further actions, such as an emergency call. The results of a preliminary evaluation of such a system based on a proof-of-concept Android app demonstrate an accurate detection of human presence within a specific range of the smartphone.