Estimation of Driver Vigilance Status Using Real-Time Facial Expression and Deep Learning

Reza Tamanani; Radu Muresan; Arafat Al-Dweik

doi:10.1109/LSENS.2021.3070419

Estimation of Driver Vigilance Status Using Real-Time Facial Expression and Deep Learning

Reza Tamanani
, Radu Muresan
, Arafat Al-Dweik

Electrical Engineering

University of Guelph

Research output: Contribution to journal › Article › peer-review

59 Scopus citations

Abstract

Drowsiness is responsible for many fatal accidents on highways. Accuracy and performance are key metrics related to many researched techniques for the detection of drivers' drowsiness. To improve these metrics, a new driver's vigilance detection system based on deep learning is proposed based on facial region diagnosis using the Haar-cascade method and convolutional neural network for drowsiness detection. Evaluation analysis of the proposed system on the University of Texas at Arlington-Real-Life Drowsiness Dataset (UTA-RLDD) dataset with stratified five-fold cross-validation showed a high accuracy of 96.8% at a speed of 8.4 frames per second, which is higher than most algorithms previously reported in the literature. For further investigation, a custom dataset including ten participants in different light conditions was collected. The conducted experiments showed the great potential of the proposed system for practical applications in intelligent transportation systems.

Original language	British English
Article number	9394715
Journal	IEEE Sensors Letters
Volume	5
Issue number	5
DOIs	https://doi.org/10.1109/LSENS.2021.3070419
State	Published - May 2021

Keywords

behavioral features
computer vision
convolutional neural network (CNN)
drowsiness detection
haar method
real-time classifier
Sensor applications

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/LSENS.2021.3070419

Cite this

@article{aa64d8ab07aa48889eccc44f995e9c32,

title = "Estimation of Driver Vigilance Status Using Real-Time Facial Expression and Deep Learning",

abstract = "Drowsiness is responsible for many fatal accidents on highways. Accuracy and performance are key metrics related to many researched techniques for the detection of drivers' drowsiness. To improve these metrics, a new driver's vigilance detection system based on deep learning is proposed based on facial region diagnosis using the Haar-cascade method and convolutional neural network for drowsiness detection. Evaluation analysis of the proposed system on the University of Texas at Arlington-Real-Life Drowsiness Dataset (UTA-RLDD) dataset with stratified five-fold cross-validation showed a high accuracy of 96.8\% at a speed of 8.4 frames per second, which is higher than most algorithms previously reported in the literature. For further investigation, a custom dataset including ten participants in different light conditions was collected. The conducted experiments showed the great potential of the proposed system for practical applications in intelligent transportation systems.",

keywords = "behavioral features, computer vision, convolutional neural network (CNN), drowsiness detection, haar method, real-time classifier, Sensor applications",

author = "Reza Tamanani and Radu Muresan and Arafat Al-Dweik",

note = "Funding Information: The work of Arafat Al-Dweik was supported by Khalifa University Competitive Internal Research Awards under Grant CIRA-2020-056. Publisher Copyright: {\textcopyright} 2017 IEEE.",

year = "2021",

month = may,

doi = "10.1109/LSENS.2021.3070419",

language = "British English",

volume = "5",

journal = "IEEE Sensors Letters",

issn = "2475-1472",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "5",

}

TY - JOUR

T1 - Estimation of Driver Vigilance Status Using Real-Time Facial Expression and Deep Learning

AU - Tamanani, Reza

AU - Muresan, Radu

AU - Al-Dweik, Arafat

PY - 2021/5

Y1 - 2021/5

N2 - Drowsiness is responsible for many fatal accidents on highways. Accuracy and performance are key metrics related to many researched techniques for the detection of drivers' drowsiness. To improve these metrics, a new driver's vigilance detection system based on deep learning is proposed based on facial region diagnosis using the Haar-cascade method and convolutional neural network for drowsiness detection. Evaluation analysis of the proposed system on the University of Texas at Arlington-Real-Life Drowsiness Dataset (UTA-RLDD) dataset with stratified five-fold cross-validation showed a high accuracy of 96.8% at a speed of 8.4 frames per second, which is higher than most algorithms previously reported in the literature. For further investigation, a custom dataset including ten participants in different light conditions was collected. The conducted experiments showed the great potential of the proposed system for practical applications in intelligent transportation systems.

AB - Drowsiness is responsible for many fatal accidents on highways. Accuracy and performance are key metrics related to many researched techniques for the detection of drivers' drowsiness. To improve these metrics, a new driver's vigilance detection system based on deep learning is proposed based on facial region diagnosis using the Haar-cascade method and convolutional neural network for drowsiness detection. Evaluation analysis of the proposed system on the University of Texas at Arlington-Real-Life Drowsiness Dataset (UTA-RLDD) dataset with stratified five-fold cross-validation showed a high accuracy of 96.8% at a speed of 8.4 frames per second, which is higher than most algorithms previously reported in the literature. For further investigation, a custom dataset including ten participants in different light conditions was collected. The conducted experiments showed the great potential of the proposed system for practical applications in intelligent transportation systems.

KW - behavioral features

KW - computer vision

KW - convolutional neural network (CNN)

KW - drowsiness detection

KW - haar method

KW - real-time classifier

KW - Sensor applications

UR - https://www.scopus.com/pages/publications/85103760375

U2 - 10.1109/LSENS.2021.3070419

DO - 10.1109/LSENS.2021.3070419

M3 - Article

AN - SCOPUS:85103760375

SN - 2475-1472

VL - 5

JO - IEEE Sensors Letters

JF - IEEE Sensors Letters

IS - 5

M1 - 9394715

ER -

Estimation of Driver Vigilance Status Using Real-Time Facial Expression and Deep Learning

Abstract

Keywords

UN SDGs

Access to Document

Other files and links

Fingerprint

Cite this