Development of a Software to Drowsiness Detection for Drivers Using Image Processing and Neural Networks

Ali   Askari; Ali Salehi  Sahlabadi; Maliheh  Eshaghzadeh; Mohsen  Poursadeghiyan; Gebraeil Nasl  Saraji

doi:10.18502/ijph.v54i9.19869

Ali Askari Department of Occupational Health Engineering, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran
Ali Salehi Sahlabadi Safety Promotion and Injury Prevention Research Center, Research Institute for Health Sciences and Environment, Shahid Beheshti University of Medical Sciences, Tehran, Iran
Maliheh Eshaghzadeh Department of Nursing, School of Nursing and Midwifery, Torbat Heydariyeh University of Medical Sciences, Torbat Heydariyeh, Iran
Mohsen Poursadeghiyan Social Determinants of Health Research Center, School of Health, Ardabil University of Medical Sciences, Ardabil, Iran
Gebraeil Nasl Saraji Department of Occupational Health Engineering, School of Public Health, Tehran University of Medical Sciences, Tehran, Iran

DOI: https://doi.org/10.18502/ijph.v54i9.19869

Keywords: Driver monitoring system; Software drowsiness detection; Neural network; Viola-Jones algorithm; Image processing

Abstract

Background: During driving, drowsiness may happen for a few moments, but its consequences can be terrible. Drowsiness in the driver can be detected in the early stages. Each method used for detecting drowsiness has its own strengths and weaknesses or benefits and flaws. The main contribution of our research was improving Driver Drowsiness Detection (D.D.D) systems.

Methods: In accordance with the research objective, it is imperative to address the subsequent inquiries (Q) throughout the process of constructing, testing, and delivering the ultimate D.D.D software model: Q1. What is the methodology employed for constructing the initial model of drowsiness detection software? Q2. How is the initial model of drowsiness detection software tested and refined during the development phase? Q3. What is the operational mechanism of the final model of drowsiness detection software?

Results: The results were able to detect different facial conditions (with hair and glasses) with a 92.3 percentage detection rate.

Conclusion: This model could help improve D.D.D systems, and detect drowsiness in different environments and situations.