AI-Driven rPPG Heart Rate Detection for In-Vehicle Monitoring

Martina Pierri, Alessandro Gumiero, Nicola Picozzi

Abstract: Road accidents are frequently attributed to driver conditions such as stress, fatigue, and drowsiness, all of which can be detected by heart rate (HR) monitoring. Remote photoplethysmography (rPPG) emerges as a promising non-invasive method that leverages standard camera technology to estimate HR by analyzing subtle skin color changes. In this article, we present a deep learning (DL) model capable of extracting HR from the rPPG temporal structure found in the public UBFC-rPPG dataset. The knowledge gained from analyzing this dataset, along with the absence of a database that reflects the dynamic conditions within a vehicle, prompted the creation of a novel Infrared (IR) dataset.

External IDs:dblp:conf/ieeesensors/PierriGP24