DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA

According to the National Highway Traffic Safety Administration, an estimated 17.6% of all fatal crashes in the years 2017–2021 involved a drowsy driver. This study proposes a drowsy driver detection system that uses both facial recognition and vehicular data to detect if a driver is feeling sleepy...

Full description

Saved in:
Bibliographic Details
Main Authors: Nur Iman Kamila Azharudin, Hafizah Mansor, Shaila Sharmin
Format: Article
Language:English
Published: UiTM Press 2024-10-01
Series:Malaysian Journal of Computing
Subjects:
Online Access:https://ir.uitm.edu.my/id/eprint/105182
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846164205774831616
author Nur Iman Kamila Azharudin
Hafizah Mansor
Shaila Sharmin
author_facet Nur Iman Kamila Azharudin
Hafizah Mansor
Shaila Sharmin
author_sort Nur Iman Kamila Azharudin
collection DOAJ
description According to the National Highway Traffic Safety Administration, an estimated 17.6% of all fatal crashes in the years 2017–2021 involved a drowsy driver. This study proposes a drowsy driver detection system that uses both facial recognition and vehicular data to detect if a driver is feeling sleepy behind the wheel. We aim to address a lack of works in the literature that combine data measured from the driver (image or biological data) and vehicular data for drowsy driver detection. Our primary data was collected from simulated driving sessions in which a camera was used to record test drivers’ faces while driving a virtual car in the CARLA simulator in both drowsy and non-drowsy states. The collected data consists of video of test drivers' faces from the camera and vehicular data from the simulated car. The video data was used to obtain facial features such as Mouth Over Eyes (MOE), Eyes Aspect Ratio (EAR), and Mouth Aspect Ratio (MAR), while the vehicle data yielded features such as speed, steering wheel movement and pedal readings. These features were used to train Support Vector Machine (SVM) and Random Forest (RF) models to detect drowsy drivers. The results indicate that RF is a better model to be used as compared to SVM in predictions of drowsiness in drivers with an accuracy of 96.24% and 86.85% respectively
format Article
id doaj-art-cbc72bdf18124e65bf35da0f130a541d
institution Kabale University
issn 2600-8238
language English
publishDate 2024-10-01
publisher UiTM Press
record_format Article
series Malaysian Journal of Computing
spelling doaj-art-cbc72bdf18124e65bf35da0f130a541d2024-11-18T14:00:03ZengUiTM PressMalaysian Journal of Computing2600-82382024-10-01921852186610.24191/mjoc.v9i2.25694DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATANur Iman Kamila Azharudin0Hafizah Mansor1Shaila Sharmin2Department of Computer Science, Kulliyyah of ICT, International Islamic University MalaysiaDepartment of Computer Science, Kulliyyah of ICT, International Islamic University MalaysiaDepartment of Computer Science, Kulliyyah of ICT, International Islamic University MalaysiaAccording to the National Highway Traffic Safety Administration, an estimated 17.6% of all fatal crashes in the years 2017–2021 involved a drowsy driver. This study proposes a drowsy driver detection system that uses both facial recognition and vehicular data to detect if a driver is feeling sleepy behind the wheel. We aim to address a lack of works in the literature that combine data measured from the driver (image or biological data) and vehicular data for drowsy driver detection. Our primary data was collected from simulated driving sessions in which a camera was used to record test drivers’ faces while driving a virtual car in the CARLA simulator in both drowsy and non-drowsy states. The collected data consists of video of test drivers' faces from the camera and vehicular data from the simulated car. The video data was used to obtain facial features such as Mouth Over Eyes (MOE), Eyes Aspect Ratio (EAR), and Mouth Aspect Ratio (MAR), while the vehicle data yielded features such as speed, steering wheel movement and pedal readings. These features were used to train Support Vector Machine (SVM) and Random Forest (RF) models to detect drowsy drivers. The results indicate that RF is a better model to be used as compared to SVM in predictions of drowsiness in drivers with an accuracy of 96.24% and 86.85% respectivelyhttps://ir.uitm.edu.my/id/eprint/105182driving datadrowsy driver detectionfacial recognitionmachine learning technique
spellingShingle Nur Iman Kamila Azharudin
Hafizah Mansor
Shaila Sharmin
DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA
Malaysian Journal of Computing
driving data
drowsy driver detection
facial recognition
machine learning technique
title DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA
title_full DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA
title_fullStr DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA
title_full_unstemmed DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA
title_short DROWSY DRIVER DETECTION SYSTEM – VIA FACIAL RECOGNITION AND DRIVING DATA
title_sort drowsy driver detection system via facial recognition and driving data
topic driving data
drowsy driver detection
facial recognition
machine learning technique
url https://ir.uitm.edu.my/id/eprint/105182
work_keys_str_mv AT nurimankamilaazharudin drowsydriverdetectionsystemviafacialrecognitionanddrivingdata
AT hafizahmansor drowsydriverdetectionsystemviafacialrecognitionanddrivingdata
AT shailasharmin drowsydriverdetectionsystemviafacialrecognitionanddrivingdata