An Attention Deep Learning Framework Based Drowsiness Detection Model for Intelligent Transportation System

Abstract

Driver drowsiness is a major cause of road accidents, making its timely detection essential for enhancing safety in intelligent transportation systems. This paper presents an attention-based deep learning framework for accurate and real-time detection of driver fatigue. The proposed model utilizes Convolutional Neural Networks (CNNs) to automatically extract discriminative facial features such as eye closure, blinking patterns, and yawning. To further improve performance, an attention mechanism is incorporated to focus on the most relevant regions of the face, enabling the system to prioritize critical indicators of drowsiness while reducing the influence of irrelevant information.

The framework is trained and evaluated on diverse datasets to ensure robustness across varying lighting conditions, driver behaviors, and environmental scenarios. Experimental results demonstrate that the proposed model achieves higher accuracy, precision, and reliability compared to conventional machine learning and standard deep learning approaches. The system is designed for real-time implementation using video input, providing immediate alerts when drowsiness is detected. Overall, the proposed attention-based framework offers an efficient, non-intrusive, and scalable solution for improving road safety and supporting advanced intelligent transportation systems.