An Extremely Lightweight Driver Distraction Detection System Using Recurrent Neural Network for Enhanced Road Safety

Road accidents worldwide have surged due to driver distractions, leading to injuries, property loss, and fatalities. Monitoring and analyzing driver behavior during driving is crucial to mitigate accidents. This study demonstrates an inexpensive distraction detection model termed the Recurrent Neural Network (RNN) based ReLU-BiLSTM model with an attention mechanism. This model processes input images from public datasets, performs preprocessing (e.g., resizing, color format conversion), and employs a ReLU-BiLSTM-based RNN for driver distraction detection. An attention mechanism enhances focus on critical information, enabling real-time detection of distractions such as cell phone use, talking, eating, sleeping, or lack of concentration. Comparative evaluations with cutting-edge techniques exhibit superior accuracy, recall, specificity, precision, and F1 score. This research underscores the potential of the suggested model to significantly enhance road safety by providing timely warnings or taking actions to mitigate distracted driving risks.