Citation

Chakraborty, Shuvo and Islam, Arfanul and Shanto, Mohammad Naimul Islam and Hasan, Sadman and Jayed, Mohammed and Ahmed, Farhana and Shah, Mohammad Shahin and Rabby, Sorowar Mahabub and Abdul Aziz, Nor Hidayati (2026) DriveGuard: Driver Expression Recognition Using Swin Transformer and CNN Variants to Enhance Road Safety. In: 2026 IEEE 2nd International Conference on Quantum Photonics, Artificial Intelligence & Networking (QPAIN), 16-18 April 2026, Chittagong, Bangladesh.

Text 3.pdf - Published Version Restricted to Repository staff only Download (316kB)

Abstract

Fatigue and alcohol impairment are significant contributors to automobile accidents, underscoring the critical need for advanced safety measures. Developing technologies capable of detecting emotional states and fatigue is vital to preventing such incidents. Among these, facial recognition sensors in autonomous vehicles have shown promise in monitoring drivers' alertness levels. However, despite their effectiveness in controlled settings, challenges remain in adapting these technologies to real-world driving scenarios. This paper introduces a novel approach that enhances the detection of driver drowsiness and related behaviors across diverse driving conditions. The proposed method integrates Eye and Mouth Closure Status (PERCLOS), a newly designed Facial Aspect Ratio (FAR), and facial expression recognition powered by Convolutional Neural Networks (CNNs). By leveraging this combination, the system achieves robust and accurate identification of drowsiness. Among the models tested, the Swin Transformer emerged as the most effective, achieving an impressive accuracy of 92 %. This superior performance highlights the Swin Transformer's capability to handle variations in angles, lighting, and facial expressions, setting it apart from other models like CNN and ResNet50. The integrated use of PERCLOS, FAR, and Swin Transformer-based facial expression detection provides a reliable framework for enhancing driver safety in real-world applications.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	DriveGuard, Driver Expression Recognition, Swin Transformer, CNN, Road Safety
Subjects:	T Technology > TE Highway engineering. Roads and pavements
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	04 Aug 2026 03:57
Last Modified:	04 Aug 2026 03:58
URII:	http://shdl.mmu.edu.my/id/eprint/16475

Downloads

Edit (login required)