Authors
College of Education, Computer Sciences, Mustansiriyah University, Baghdad, Iraq
Abstract
Driver fatigue and distraction are among the leading causes of road accidents worldwide, posing a critical challenge for intelligent transportation systems (ITS). Vision-based driver monitoring has emerged as a promising solution due to its non-intrusive nature and compatibility with in-vehicle environments. However, many existing approaches rely on isolated behavioural cues, lack real-time capability, or require specialized sensing hardware, which limits their practical deployment. This paper presents a real-time vision-based framework for monitoring driver fatigue and distraction using a single in-cabin RGB camera. The proposed framework integrates complementary behavioural indicators, including eye closure analysis based on PERCLOS, yawning detection, and head pose estimation, through a decision-level fusion strategy. The framework is evaluated on two widely adopted public benchmark datasets, namely the NTHU Drowsy Driver Detection dataset and the YAWDD dataset. Experimental results show that the proposed approach achieves detection accuracies of 92.6% on NTHU and 93.0% on YAWDD, with average processing latencies below 120 ms, satisfying real-time operational requirements. These results demonstrate that multi-cue visual analysis significantly improves detection robustness compared to single-cue methods while maintaining practical deployability. The proposed framework therefore provides an effective and scalable solution for real-time driver state monitoring and contributes to enhanced road safety in intelligent transportation systems.