Volume 6, Issue 3, March – 2021 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Drowsy Driving: A Look at the Monitoring and

Detection System
Okeke, Ogochukwu C.*1, Okonkwo Chinonso J*2
1
* Department of Computer Science, Chukwuemeka Odumegwu Ojukwu University, Uli, Anambra State.
*2 Department of Computer Science, Chukwuemeka Odumegwu Ojukwu University, Uli, Anambra State

Abstract:- The rate at which cars and other motor Other steps towards solving this problem is the
vehicle related accidents have increased in recent years physiological approach which employs electro-cardiological
is becoming mind bulging and fearful and majority of signal collected from the driver for monitoring.
these mishaps are drowsy driver related. This has
resulted in the loss of lives, goods and properties etc. The The other approach is the behavioral approach which
essence of this analysis is to study and review the collected data from the driver using image acquisition tool
previous works on drowsy driving related accidents, and the process using various image processing technique
their causes and measures taken. The gaps in these like in (Barr et al., 2019) and artificial intelligence technique
studies were noted in order to propose and design a new (Metin Kaya, 2018), however these approaches suffers from
or a robust system. This was achieved using various lack of adaptive intelligence.
techniques such as image acquisition, computer vision,
face detection, feature extraction, training and This research work therefore proposes a new technique
classification. The techniques were designed using to solve this problem using computer vision based artificial
universal modeling diagram and mathematical modeling intelligence technique.
approach based on the requirements for object oriented
analysis design methodology adopted for the study. The  Statement of the Problem
designs were implemented as a prototype system using (a) High rate of false alarm prompting in the conventional
MATHLAB and tested with real time driving behaviors. system due to combination of all attributes drowsiness in
one dataset.
Keywords:- Drowsy-Driving, Mathlab, Computer Vision, (b) Poor data quality due to lack of intelligence by existing
Face Detection, Image Acquisition. image acquisition system.

I. INTRODUCTION  Aim and Objectives of the Study

The aim of this research is the development of a
 Background to the Study drowsy driver monitoring and detection system with the
In the World Bank report of 2014, it was estimated following setout objectives;
that over 900 million people are being transported with their i. To design a drowsy driver dataset using universal
goods and services every day by vehicles all over the world modeling approach in order to pinpoint the exactness of
(World Bank, 2014). Majority of these people depends on whether a driver is drowsy or not before the alarm comes
public transport for this locomotive process, to perform up.
various daily routines like going to work, schools, service ii. To develop a Computer Vision based algorithm for
delivery system among other day to day activities. tracking of drivers’ behavior attributes.

Drowsy driving is the operation of a motor vehicle II. LITERATURE REVIEW

while cognitively impaired by lack of sleep (Wikipedia
project, 2020). This lack of sleep is the major cause of Driving is a complex task that requires the driver’s full
drowsiness with 23% of all adult drivers feeling sleep while attention. When the driver is not completely engaged within
driving at least once per month in the last five years this task, many factors can reduce his/her ability to detect
(National sleep foundation, 2014). Drowsiness in driving and to react to risky situations. These factors which
has resulted to a lot of road catastrophes involving loss of contribute to the nonlinear physiological state of a driver
loved ones, friends, families, properties (vehicle) among includes stress, fatigue, spending much time on the steering
others. driving, hunger, lack of sleep, night driving which all can
cause physiological state of drowsiness in drivers.
To solve this problem, various administrative steps
have been proposed and implemented over time such as the Drowsiness can be defined as the state when one has a
federal high way road safety commission, motor traffic strong desire for sleep or sleeping unusually for a long
division of the police sector among other agencies which period of time. It is associated with reduction in response
have all established and put into law various rules and time gradually, processing a less vigilant behavior, inability
regulations to help control accident on the high ways. to process available information promptly and errors in

IJISRT21MAR588 www.ijisrt.com 969

Volume 6, Issue 3, March – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
short-term memory according to Ahmad et al, (2019). This Facial Features. The research applied Viola Jones algorithm
state of drowsiness can also be associated with stress and for skin colour pixels detection, detection of the position for
fatigue that limits the performance of an individual eyes and mouth, then the data threshold collected were
unexpectedly. calculated for result prediction.

Drowsiness is a dangerous state of physiology that has Wei et al, (2017), researched on Driver Drowsiness
been proven to have significant impact on the passengers, Recognition Based on Computer Vision Technology. They
drivers and road safety, reflecting driver’s unsafe behavior. focused on eliminating nonlinearity on image captured due
In a research conducted by Williamson et al. (2014) using to illumination and driver posture. They used eye-tracking
driving simulator on drowsy driving revealed that higher and image processing algorithms like Fisher’s linear
levels of sleepiness were more likely to be involved in road discriminant functions to measure some the nonlinear
crashes and centerline crossings. Furthermore, they revealed features based on stepwise method to reduce co-relations
that drowsy drivers have slower response to sudden and then extract an independent index which was used for
nonlinear vehicle dynamics. drowsy recognition. Ashish and Rusha (2018) researched on
Driver Drowsiness Monitoring System using Visual
Drowsiness is a major contributor to most of the road Behaviour and Machine Learning, they acquired real time
disasters happening everyday which results to numerous data for analysis using webcam that records the face of the
injuries, deaths and economic cost to the society. In the driver and then process the data captured with image
United States for instance, it was reported by the National processing techniques. Machine learning was used to train
Highway Traffic Safety Association (2017) that 795 people and then predict the visual behaviour of the driver.
deaths and over 90,000 car crashes occurred.
From the review it was revealed that the two main
Over time, various researches have been done to approach which are physiological and behavioral based
identify the main causes of drowsiness during driving. Ting approaches having used in the past to solve this problem of
et al. (2018) identified time on task (increased time on the drowsy driver monitoring and detection despite their success
steer wheel driving) as the most well known cause of driver have recorded many constraints such as false alarm, design
drowsiness. This was justified by other studies which complexity, poor detection and prediction accuracy,
revealed that increase driving time with increasing unreliability, high cost among others. These challenges have
subjective sleepiness and fatigue, increasing steering wheel undermined the performance of the existing system as
movement and longer blink durations all result to identified in the empirical review, hence there is need for an
drowsiness during driving (Morales et al., 2017) revealed intelligent system which is adaptive to localize and collect
that deprivation of sleep is a strong factor that clearly drivers behavior in real time and then use the data collected
induces higher rate of sleepiness and opined that even mild to classify drowsiness with high accuracy using artificial
levels of sleep deprivation of two hours can impair the intelligence technique.
performance of driver on the steering. In the same vein, the
research suggests that temperature, light settings,  Proposed System and Implementation.
geometry’s monotony, traffic conditions, are some of the The proposed system will be developed using
external factors which can influence drowsiness in real time. computer vision based artificial intelligence technique to
help address the challenges identified in the existing system.
In the year 2018, Taner et al., published a paper on From the existing system analysis it was observed that the
Drowsy Driver Detection System using Eye Blinking accuracy of the system is highly dependent on the quality of
Patterns. The horizontal symmetry feature of the eyes was data collection and also the intelligence of the classification
used to detect the visual changes in the eye locations. The approach, hence the research proposed an adaptive image
detection was performed in real time with the use of a acquisition algorithm which will ensure intelligent data
standard webcam with high resolutions and aspect ratios to collection of the driver’s behavior and classify drowsiness.
detect eye blinks and then result was predicted from ground
truth driver image using similarity differential model. Marco The artificial intelligence technique will be trained
et al, (2015) researched on Driver Drowsiness Warning using a new data model which will be designed using data
System Using Visual Information for Both Diurnal and containing various drowsy attributes such as eye closed
Nocturnal Illumination Conditions. A new module for (sleep), yawning, continuous eye blinking and drivers head
Advanced Driver Assistance System (ADAS) was down. This data will be collected in video format and used
introduced which uses facial visual information to detect to develop the training dataset to learn the artificial
driver drowsiness based on already trained Artificial intelligence technique of the drowsy intelligence as a
Intelligence. The real-time system works in different light drowsy reference model. Other process and functionalities
conditions (day and night driving). Ijaz (2014) presented a of the proposed system are portrayed in the proposed
research on Driver’s Fatigue Detection System Based on diagram below;

IJISRT21MAR588 www.ijisrt.com 970

Volume 6, Issue 3, March – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165

Fig 1:- The proposed system

 System implementation computer vision is used for face tracking and detection. The
The system is implemented using the models data collected is converted using the activation function
developed, image acquisition toolbox, computer vision based on the statistics toolbox and feed forward for training
toolbox, neural network toolbox, statistics and machine with the neural network training tool to classify the desired
learning toolbox and Mathlab. The image acquisition result.
toolbox is used to activate the hardware device, and then the

Fig 2:- system implementation

IJISRT21MAR588 www.ijisrt.com 971

Volume 6, Issue 3, March – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
III. METHODOLOGY [7]. Williamson, A.; Friswell, R.; Olivier, J.; Grzebieta, R.
Are drivers aware of sleepiness and increasing crash
This section discussed the methodology employed for risk while driving? Accid. Anal. Prev. 2014, 70, 225–
the development of the new system. This was developed 234.
based on object oriented analysis design methodology [8]. National Highway Safety Administration Report
(OOADM). This methodology was employed to bring out (NHSAR), United Sates, 2016
detailed description of the system as well as providing [9]. Ting, P.-H.; Hwang, J.-R.; Doong, J.-L.; Jeng, M.-C.
avenue for easy modification of the system as need may Driver fatigue and highway driving: A simulator study.
arise in future. OOADM suits in analyzing and designing Physiol. Behav. 2018, 94, 448–453.
large system like the case study as it gives out a clearer view [10]. Morales, J.M.; Díaz-Piedra, C.; Rieiro, H.; Roca-
of the modules, procedures, functions and sub-systems González, J.; Romero, S.; Catena, A.; Fuentes, L.J.; Di
alongside with their respective relationships. This also gives Stasi, L.L. Monitoring driver fatigue using a single-
a clearer picture and the representation of the objects (data channel electroencephalographic device: A validation
and processes) as contained in the drowsy driver software as study by gaze-based, driving performance, and
such giving the designers a complete analysis for the subjective data. Accid. Anal. Prev. 2017, 109, 62–69
development of efficient system that meet specifications as [11]. Wei Zhang, Bo Cheng, Yingzi Lin (2017). Driver
contained in the specification documents and provides Drowsiness Recognition Based on Computer Vision
design with ease in modification. Technology. TSINGHUA SCIENCE AND
TECHNOLOGY ISSNll1007-0214ll18/18llpp354-362
IV. CONCLUSION Volume 17, Number 3, June 2012.
[12]. IJAZ KHAN (2014). DRIVER’S FATIGUE
This study has successfully developed a drowsy driver DETECTION SYSTEM BASED ON FACIAL
monitoring and detection system. The research designs a FEATURES. Degree of Electrical and Electronics
standard drowsy dataset, intelligent video based drowsy data Engineering, Faculty of Electrical and Electronics
collection system using computer vision and also an Engineering Universiti Tun Hussein Onn Malaysia.
intelligent classification system using artificial neural
network. The systems were designed using universal
modeling diagram and mathematical models and then
implemented with MATHLAB programming editor as a
prototype system. The system was tested and the result
shows correct driver behavior classification accuracy of
98.6%.

REFERENCES

[1]. World bank funded research on transportation and

road accident, 2014 (Accessed August, 31st , 2020).
[2]. National Sleep Foundation, (2014). Facts and Stats,
http://drowsydriving.org/about/facts-and-stats/.
Accessed on January 30th
[3]. Barr L., Howrah H., S. Popkin, RJ. Carrol l(2019) “ A
review and evaluation of emerging driver fatigue
detection, In Proceedings of the IEEE International
Conference on Intelligent Transportation Systems,
Maui, Hawaii, HI, USA, 5–8 November 2019; pp.
1394–1402.
[4]. Marco Javier Flores, José María Armingol and Arturo
de la Escalera, (2014). Real-Time Drowsiness
Detection System for an Intelligent Vehicle. Intelligent
Systems Laboratory at Department of System
Engineering and Automation, Universidad Carlos III
de Madrid, Leganés (Madrid) Spain.
[5]. Metin Kaya, ”Image Clustering and Compression
Using an Annealed Fuzzy Hopfield Neural Network”,
2018
[6]. Ahmad Kamran, Malik M. Naeem Mannan and
Myung Yung Jeong, (2019). Drowsiness, Fatigue and
Poor Sleep’s Causes and Detection: A Comprehensive
Study. Pusan National University, Busan, South
Korea. (No.2017R1A2B2006999). DOI
10.1109/ACCESS.2019.2951028.

IJISRT21MAR588 www.ijisrt.com 972