REAL TIME DRIVER’S CONSCIOUSNESS

DETECTOR
Saptarshi Bhattacharya Sudhanshu Ranjan Dr.P.Nancy
Dept. of Computing Technologies Dept. of Computing Technologies Faculty of Engineering and Technology
SRM Institute of Science and Technology SRM Institute of Science and Technology SRM Institute of Science and Technology
Kattankulathur, Tamil Nadu-603203, Kattankulathur, Tamil Nadu-603203, Kattankulathur, Tamil Nadu-603203,
India India India

Abstract—Drowsy or drunk driving poses significant risks to injuries and deaths is rising.globally every year. Drunk driving
drivers, passengers, and pedestrians. A research paper explores poses a serious danger to both drivers and pedestrians. Every
using physiological indicators like eye size, eye openness, and year, about 1.25 million people die in auto accidents. For
head movements to detect driver consciousness levels and signs of
drunkenness. Real-time data from onboard cameras is analysed those between the ages of 15 and 29, traffic accidents are the
by the research using machine learning methods and enhanced main cause of mortality [25]. Almost forty percent of traffic
image processing. The study article explores physiological indica- accidents are the result of drunk driving [26]. Over the last
tions as well as the creation of an all-encompassing framework 10 years, a number of image processing-based methods for
for real-time assessment of driver impairment. This paradigm recognising intoxicated drivers have been developed and put
takes into account contextual aspects including driving patterns,
vehicle dynamics, and ambient circumstances in addition to the into use. The study’s findings have prompted the suggestion
examination of individual physiological signs. The objective of of a strategy to reduce the number of accidents. The approach
this study is to improve the precision and dependability of identi- primarily focuses on head motions and the ratios of yawning
fying intoxication or sleepiness by including these complex factors to eye closure. Things may not be right if the driver seems to
into the analysis. The research also looks at using sophisticated be dozing off or intoxicated.
feature extraction and fusion techniques, which would allow the
system to detect minute behavioural changes in drivers that might This research used a technique known as ”eye localization,”
be signs of impairment. A coherent picture of the driving state which comprises analysing a face picture and identifying
is created by combining pertinent characteristics extracted from the eyes using the Matlab programme. After establishing the
the data streams using techniques like sensor fusion and deep subject’s location, the apparatus can ascertain whether or not
learning. Thanks to this all-encompassing strategy, the system
their eyes are open. It could also be able to detect fatigue
can adjust to various driving conditions and distinguish between
typical fluctuations and indications of impairment. indicators. The goal of this research is to identify fatigued
The study also tackles issues like computing efficiency, re- driving in order to improve traffic safety and reduce collisions.
silience against noise in the environment, and privacy problems Using a camera aimed straight at the driver, one technique to
that arise when implementing such systems in the real world. The gauge driver weariness is to capture a real-time image of the
study intends to make it easier to integrate these technologies into
driver’s face. When the video is finished, watch for indications
current automobile platforms without incurring a large overhead
by creating lightweight algorithms and optimising computing of fatigue in the subject’s eyes and face. The head motions of
resources. the driver are also observed.
The gadget monitors your eyes to detect the condition of
I. I NTRODUCTION your eyelids, including whether they are open or closed. In
The face is a crucial physical characteristic that conveys a order to alert drivers to the situation when drunkenness is
great deal of information. The facial expressions of a regular detected, a warning light is activated. The fraction of the
motorist differ from those of a weary driver. For example, whole time that an eye is closed at any specific point may be
a tired driver may yawn and blink more often than normal. expressed. When the eyelid closures happen more often than
Even if driverless cars will soon become mainstream, human a preset threshold, the system recognises driver weariness. An
drivers will still be needed for some time to come. Studies alarm will thus sound to wake them.
have shown that alcohol use results in observable changes to
the patterns of eye movement and emotions on the face[24]. II. L ITERATURE S URVEY
Drinking alcohol is also linked to fatigue and sleepiness. It
is possible to evaluate these variances by looking at images This study looked at a technique for identifying variations in
of people who are both sober and drunk. Being aware of the a driver’s level of eye openness. There is a strong correlation
warning symptoms of weariness and exhaustion is crucial to between eye-opening variations and fatigue. This method finds
reducing your risk of being involved in an accident, since alterations in the physical characteristics of physiological func-
driver drowsiness is one of the most common causes of tions. This strategy was chosen because early identification at
accidents on today’s roadways. Consequently, the number of the onset and a high degree of detection accuracy comparable
to methods based on physiological indicators would be nec- possible uses. The authors assess a wide range of sensor-based
essary for a functional drowsiness detection system. This was techniques, such as breath analyzers, wearable biosensors, bio-
one of the factors in the decision to use this tactic. The device chemical assays, and computer vision systems, through a thor-
must also have the ability to identify driver fatigue through a ough examination of the literature. Each sensing technique’s
noncontact approach. guiding principles are covered in the article, with special atten-
In order to reduce accidents worldwide, we are currently de- tion paid to each one’s sensitivity, specificity, response time,
veloping an intelligent system that can correctly identify both and mobility. The authors also examine the practical aspects
drunkenness and sleepiness. In order to conduct this study, of ethanol detection, including cost-effectiveness, regulatory
a real-time smartphone application would assess a subject’s compliance, and calibration needs. The paper provides useful
awareness by analyzing their facial patterns. (Whether the insights into developing and deploying alcohol detection tech-
individual is inebriated from alcohol, sleepy from alcohol, or nologies for various settings, such as transportation, workplace
both) The following are the benefits of our suggested system safety, and law enforcement, by synthesizing the current state-
over the current one: The system continuously monitors the of-the-art in ethanol intoxication sensing.
user’s consciousness and allows real-time detection; it detects This offers a novel method for identifying alcohol in-
fatigue or intoxication and sounds an alarm to alert the user; toxication by employing neural networks to analyze speech
The previous method was less secure than this one since it was patterns. The study explores the connection between drinking
more straightforward to identify drunkenness and drowsiness. alcohol and changes in speech traits such as pitch variation,
The program will be easier to use if the detecting systems articulation clarity, and speech pace. Based on these attributes,
are combined into a single application. Both Studies A and the research suggests a methodology for automatically identi-
B (Van Dongen et al., 2011a and 2011b) were carried out fying drunken speech using improvements in machine learn-
in a laboratory setting using a high-fidelity driving simulator ing, namely neural network models. By utilizing supervised
(Van Dongen et al., 2010). Since the designs of these studies learning algorithms and feature extraction approaches, the
are similar, we will discuss them all together in this section. system gains the ability to accurately distinguish between
Eyelid factors and support vector machines are used to detect speech samples that are drunk and those that are not. The
driver tiredness. The EOG experimental data is used to derive study also addresses the usefulness of speech-based intoxica-
the eyelid movement characteristics required for SVM training tion detection in a variety of settings, such as public health
and validation. The initial step involves locating eye blinks in campaigns, occupational safety, and law enforcement. The re-
the EOG data. The blinks are preserved when the two EOG search advances the creation of non-intrusive, real-time alcohol
vertical channels are combined into one channel during the intoxication detection systems that can improve safety and
averaging process. 95 percent of the blinks could be reliably reduce the hazards related to impaired behavior by combining
identified after the combined signal was analysed. This amount insights from speech analysis and artificial intelligence.
was thought to be adequate for the investigation carried out in
the section that follows. III. M ETHODOLOGY
This device uses an infrared camera to assess the driver’s
condition in both daylight and nighttime conditions. Symmetry We use a simple general pattern to determine if someone is
is used as the fundamental idea in the recognition of faces tired or not.
and eyes. The choice of eye states is made using the Hough 1. A moving video feed is initially set up on a camera to
Transform for Circles. look for faces.
We looked at three driving datasets, each of which included 2. After locating a face, the eye regions could be retrieved
drivers with varying degrees of sleep deprivation and classified using facial landmark detection.
every minute of driving as fatigued or awake. HRV served 3. The regions around the eyes are inspected in order to
as the basis for two different sleepiness detectors that we establish the eye aspect ratio, which is then utilised to assess
developed. A driver’s fitness to drive was determined by whether or not the eyes are closed.
the sleep-deprivation detector, while the drowsiness episodes 4. The alarm is activated to wake the driver awake if the
detector classified each minute of driving as ”awake” or eye aspect ratio shows that they have been sleeping with their
”drowsy” based on seven HRV-derived parameters (positive eyes closed for an extended period of time. The pillion drivers
predictive value 0.96, sensitivity 0.59, and specificity 0.98 on are also notified by the indicator lights. In this instance, the
3475 min of driving). This category was decided by how little eyelids had remained closed long enough.
sleep the driver had had before the driving session began. Face Recognition:- First, the suggested approach will take
In order to establish if a participant was experiencing sleep the form of taking individual video frames. The image that
deprivation, just one HRV component was used for the first best captures the person’s face will be chosen by the computer.
three minutes of driving (positive predictive:-0.80, sensitivity:- The HAAR face detection algorithm is used to complete this
0.62, specificity 0.88 on 30 drivers). challenge properly. The cascade file has to load before the
The study offers a thorough analysis of the technologies resultant frame can be forwarded to an edge identification
and procedures now in use for ethanol intoxication detection, algorithm that looks for possible objects of various sizes. Set
with an emphasis on their advantages, disadvantages, and the edge detector to only detect objects that fit into a specific
size range, like the face region, rather than trying to detect IV. C ONCLUSION
features at any size. In summary, the integration of head movement analysis,
HAAR Cascade Classifier:- It was among the first meth- intoxication detection, and tiredness detection with the driver
ods of recognizing a face, especially a driver’s. One of the monitoring system aims to enhance road safety and mitigate
few methods for object identification combined with facial the potential hazards associated with intoxicated or sleep-
recognition. Paul Viola and Michael Jones came up with deprived drivers. The goal of this system is to avoid accidents
this scheme. It gains knowledge from hundreds of images of by analysing driver behaviour in real-time and delivering
various people’s faces taken in various lighting conditions. The timely warnings and interventions via the use of machine
validation process for the HAAR Cascade Classifier involved learning algorithms and powerful computer vision methods.
ten participants. The process of detecting impairments is facilitated by head
Region of Interest:- A zone of interest (ROI) can be used movement analysis, which records additional indications such
to obtain a driver’s face recognition more accurately. T= head orientation, tilt, and movement speed. In order to facil-
(100-—FC/2—)/100 is the equation for changing temperature, itate proactive intervention, intoxication detection algorithms
where FC is the face degree and T is the temperature. The categorise both sober and inebriated head motions. By linking
angles created when two parallel lines intersect are known as differences in head movement patterns to tiredness-related
face degrees. symptoms, fatigue detection enhances the system’s capacity
Eye Detection:- The eyes identification function looks for to recognise sleepiness indicators. The technology can save
the driver’s eyes after the face detection function correctly lives and greatly increase road safety, but further study and
recognizes the driver’s face. Once the face has been identified, development is required to enhance its use and performance
the eye region needs to be found. Keep in mind that the eyes in a variety of driving scenarios and demographics.
are limited to the upper portion of the face. You can see the
R EFERENCES
Region of Interest (ROI) around the eyes after removing the
lips and hair from the upper edge of the face. Results are [1] Sahayadhas, Arun, Kenneth Sundaraj, and Murugappan Murugappan.
”Detecting driver drowsiness based on sensors: a review.” Sensors 12.12
more accurate when the processing speed is increased flow and (2012): 16937-16953.
position estimation are two examples of algorithms that can [2] Ueno, Hiroshi, Masayuki Kaneda, and Masataka Tsukino. ”Development
be used to extract useful characteristics from video footage. of drowsiness detection system.” Proceedings of VNIS’94-1994 Vehicle
Navigation and Information Systems Conference. IEEE, 1994.
Eye State and Drowsiness Detection Finding out if the [3] Ramzan, Muhammad, et al. ”A survey on state-of-the-art drowsiness
person’s eyes were open or closed during the pertinent time detection techniques.” IEEE Access 7 (2019): 61904-61919.
[4] Santamaria, Joan, and Keith H. Chiappa. ”The EEG of drowsiness in
period is the first step in diagnosing sleepiness. When driving, normal adults.” Journal of clinical Neurophysiology 4.4 (1987): 327-382.
the Eye Aspect Ratio (EAR) is used to measure whether the [5] Flores, M.J., Armingol, J.M. and de la Escalera, A., 2008, June. Real-
driver’s eyes are open or closed. The Euclidean formula is time drowsiness detection system for an intelligent vehicle. In 2008
IEEE Intelligent Vehicles Symposium (pp. 637-642). IEEE.
used to calculate the EAR using the separation between the [6] Čolić, A., Marques, O. and Furht, B., 2014. Driver drowsiness detection:
critical locations on the eyes. This connection is expressed by Systems and solutions (p. 55). Cham, Switzerland: Springer International
the following equation, which is the Eye Aspect Ratio (EAR): Publishing.
[7] Siddiqui, H.U.R., Saleem, A.A., Brown, R., Bademci, B., Lee, E., Rus-
In this equation, the denominator indicates the distance tam, F. and Dudley, S., 2021. Non-invasive driver drowsiness detection
between horizontal eye landmarks and the numerator indicates system. Sensors, 21(14), p.4833.
the distance between vertical eye landmarks.A recommended [8] Yan, J.J., Kuo, H.H., Lin, Y.F. and Liao, T.L., 2016, July. Real-time
driver drowsiness detection system based on PERCLOS and grayscale
threshold of 300 has been established for the model. When image processing. In 2016 International Symposium on Computer,
the EAR falls below 300, the driver is considered tired and Consumer and Control (IS3C) (pp. 243-246). IEEE.
should not be operating a vehicle. [9] Poursadeghiyan, M., Mazloumi, A., Saraji, G.N., Baneshi, M.M., Kham-
mar, A. and Ebrahimi, M.H., 2018. Using image processing in the
Head Movement Detection: To track the driver’s head proposed drowsiness detection system design. Iranian journal of public
motions in real time, incorporate a head movement detection health, 47(9), p.1371.
[10] Ramzan, M., Khan, H.U., Awan, S.M., Ismail, A., Ilyas, M. and
module. Analyze changes in head orientation, tilt, and move- Mahmood, A., 2019. A survey on state-of-the-art drowsiness detection
ment speed using computer vision algorithms. techniques. IEEE Access, 7, pp.61904-61919.
[11] Ngxande, M., Tapamo, J.R. and Burke, M., 2017. Driver drowsiness
Fatigue detection: Use drowsiness and facial recognition detection using behavioral measures and machine learning techniques: A
with HAAR Cascade facial markers to detect fatigue. It review of state-of-art techniques. 2017 Pattern Recognition Association
identifies the temporal shift between frames in a sample and of South Africa and Robotics and Mechatronics (PRASA-RobMech),
pp.156-161.
extracts a discriminative representation for the aligned faces. [12] Saini, V. and Saini, R., 2014. Driver drowsiness detection system and
Real-time Alerting and Intervention: Put in place a real-time techniques: a review. International Journal of Computer Science and
Information Technologies, 5(3), pp.4245-4249.
alerting system to inform the driver and other passengers as [13] Alshaqaqi, B., Baquhaizel, A.S., Ouis, M.E.A., Boumehed, M., Ouamri,
soon as indicators of weariness or intoxication are noticed. Use A. and Keche, M., 2013, May. Driver drowsiness detection system. In
visual indicators and aural alarms to notify the driver when to 2013 8th International Workshop on Systems, Signal Processing and
their Applications (WoSSPA) (pp. 151-155). IEEE.
take a break or give over control to a co-driver, among other [14] Deng, W. and Wu, R., 2019. Real-time driver-drowsiness detection
corrective measures. system using facial features. Ieee Access, 7, pp.118727-118738.
[15] Stancin, I., Cifrek, M. and Jovic, A., 2021. A review of EEG signal
features and their application in driver drowsiness detection systems.
Sensors, 21(11), p.3786.
[16] Park, S., Pan, F., Kang, S. and Yoo, C.D., 2016, November. Driver
drowsiness detection system based on feature representation learning
using various deep networks. In Asian Conference on Computer Vision
(pp. 154-164). Springer, Cham.
[17] Mohanty, S., Hegde, S.V., Prasad, S. and Manikandan, J., 2019, Novem-
ber. Design of real-time drowsiness detection system using dlib. In
2019 IEEE International WIE Conference on Electrical and Computer
Engineering (WIECON-ECE) (pp. 1-4). IEEE.
[18] Samiee, S., Azadi, S., Kazemi, R., Nahvi, A. and Eichberger, A.,
2014. Data fusion to develop a driver drowsiness detection system with
robustness to signal loss. Sensors, 14(9), pp.17832-17847.
[19] Sałapatek, D., Dybała, J., Czapski, P. and Skalski, P., 2017.
Driver drowsiness detection systems. Zeszyty Naukowe Instytutu Po-
jazdów/Politechnika Warszawska, 3(112), pp.41-48.
[20] Hashemi, M., Mirrashid, A. and Beheshti Shirazi, A., 2020. Driver safety
development: Real-time driver drowsiness detection system based on
convolutional neural network. SN Computer Science, 1(5), pp.1-10.
[21] Mehta, S., Dadhich, S., Gumber, S. and Jadhav Bhatt, A., 2019,
February. Real-time driver drowsiness detection system using eye aspect
ratio and eye closure ratio. In Proceedings of international conference
on sustainable computing in science, technology and management (SUS-
COM), Amity University Rajasthan, Jaipur-India.
[22] Global status report on road safety 2015. 2015. World Health Organi-
zation: Geneva, Switzerland, 2015.
[23] Drinking, Driving: A Road Safety Manual for Decision-Makers, and
Practitioners. 2007. Global Road Safety Partnership, Geneva, Switzer-
land.