Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

Driver Drowsiness Detection and Alert

Generating System
Sowntharya Lakshmi B Deepika A
Dept. of Master of Computer Application Dept. of Master of Computer Application
Sona College of Technology Sona College of Technology
Tamil Nadu, India Tamil Nadu, India

Sree Madhumitha J Nandhini R

Dept. of Master of Computer Application Dept. of Master of Computer Application
Sona College of Technology Sona College of Technology
Tamil Nadu, India Tamil Nadu, India

Suhana Nafais A
Assistant Professor
Dept. of Master of Computer Application
Sona College of Technology
Tamil Nadu, India

Abstract:- Driver drowsiness detection and alert and an alert system to inform the accident area to ambulance
generating system is a form of artificial intelligence that driver. The system works by monitoring the driver's
practices machine learning algorithms to detect signs of behaviours and expressions, such as eye blinking frequency
drowsiness in drivers. This technology typically involves and yawning, and comparing it to a predetermined baseline.
sensors such as cameras, microphones, and If the driver's behaviours deviate from the baseline, the
accelerometers that track the driver’s behaviours, such system will alert the driver of their fatigue and suggest that
as head position, eye movement, and yawning, and then they take a break. For this process, we use machine learning
use machine learning algorithms to detect patterns of algorithm. We used the haar cascade algorithm The Haar
drowsiness. The technology can then alert the driver if Cascade technique is used to recognise objects in pictures. It
they are at risk of dozing off, helping to reduce the risk is a machine learning approach in which a cascade function
of accidents or dangerous driving behaviour. A machine- is taught using a large number of positive and negative
learning approach to detect drowsiness in drivers using images. Then it's used in other photos to detect things. The
facial landmarks. The planned system uses a Haar Cascade algorithm is commonly used for face
convolutional neural network (CNN) to observe the detection and recognition. It is also used for driver
facial features of a driver in real-time and then compares drowsiness detection. In driver drowsiness detection, the
them with a set of predefined features associated with Haar Cascade algorithm is used to observe the face of the
drowsiness. The system can then alert the driver of their driver and then analyze the facial expressions. If the facial
drowsiness by sounding an alarm or displaying a expressions indicate that the driver is drowsy or sleepy, an
warning message. The planned method can be used in alarm is triggered. This alarm alerts the driver and prevents
various applications, such as driver assistance systems, him/her from falling asleep while driving, thus avoiding
autonomous vehicle systems, and public safety systems. accidents. Drowsiness detection and alert -generating
Lastly, we outline the issues that current systems systems can be beneficial in reducing the number of
confront and discuss the associated research prospects. accidents caused by fatigued drivers, as well as improving
driver safety. The process for a drowsiness detection and
Keywords:- Machine Learning, Autonomous Vehicle alert system typically involves the system capturing pictures
Technology, Yawn Detection, Driver Drowsiness Detection, of the driver’s face using a camera, Computer vision
Image-Based Measures. algorithms process the images to detect and track facial
features such as the eyes, eyebrows, and mouth. The system
I. INTRODUCTION computes a baseline of the driver’s behaviours from the
facial features. The system observe the driver’s behaviours
Drowsiness detection and Alert Generating systems over time and compares it to the baseline. If the driver’s
are designed to observe signs of exhaustion and drowsiness behaviours deviates from the baseline, the system will alert
in drivers, alerting them to the fact that they need to rest or the driver of their fatigue and suggest that they take a break.
take a break from driving. A drowsiness detection system If unfortunately accident is occurred than the system also
typically consists of a camera, machine learning algorithms, can alert emergency services like ambulance if the driver got

IJISRT23MAY682 www.ijisrt.com 950

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
severe injuries. This can be done by sending a GPS location are merged using Fisher's linear discriminated functions
to the emergency services. If the driver got injuries due stepwise manner. The results of six participants in driving
drowsiness, then with help of developer API of ambulance prototype experiments show that this video-based
booking and tracking application, send location of that car to drowsiness recognition technique can be implemented with
nearby ambulance drivers to rescue the people. For this 86% accuracy.
process we use API technology. APIs are collections of
descriptions and rules that enable software components to In 2013, G. Kong et. al. described 'Visual Analysis of
communicate and interact with one another via a simple Eye State and Head Pose for Driver Alertness Monitoring'.
array of commands. APIs serve as messengers, delivering a They proved a visual examination of eye state and head pose
request from one application to another and returning an (HP) for constant monitoring of a vehicle driver's awareness.
answer in real time. Using this method only ,we are send To evaluate the driver's drowsiness or diversion level, most
location of area where the accident is occur to driver of the conventional techniques to visual identification of non-alert
nearby ambulance. driving structures depend on either eye closure or head
nodding angles. The proposed scheme obtains critical
II. LITERATURE SURVEY information on a vehicle driver's non-alertness using image
features such as eye index (EI), pupil activity (PA), and HP.
In 2008, Hong Suet.al. described ‘A Partial Least A support vector machine (SVM) identifies driving events
Squares Regression-Based Fusion Model for Predicting the as alert or non-alert based on a sequence of video frames. In
Trend in Drowsiness was proposed to manage with the real-world driving conditions, experimental outcomes
problem of strong collinear relations among eyelid indicate that the suggested scheme provides a high degree of
movement features. The predictive accuracy and strength of precision with acceptable errors as well as false alarms for
the model were validated, which showed that it provided a individuals of different ethnicities and genders.
different way of combining multi-features for enhancing our
capability of detecting and predicting the state of drowsiness In June 2014, Enosis’s et. al.described ‘Driver
’. Drowsiness Detection through HMM based Dynamic
Modelling’. They suggested an innovative approach for
In June 2010, Bin Yang et. al. described that measures detecting drowsiness by analyzing the driver's facial
of the driver's eyes can be used to detect drowsiness under a expression using Hidden Markov Model (HMM)-based
trainer or experimental conditions. The performance of the dynamic modeling. They used a driving simulation setup to
latest eye-tracking-based in-vehicle exhaustion prediction implement the algorithm. The efficiency of the suggested
measures is evaluated. These measures are evaluated approach was confirmed by experimental results.
statistically and by a classification method based on a large
dataset of 90 hours of real road drives. In August 2014, García et. al. [21] described ‘Driver
Monitoring Based on Low-Cost 3-D Sensors’. They
In 2011, M.J. Flores et. al. described a 'Driver demonstrated a method for driving surveillance and incident
drowsiness detection system under infrared illumination for detection method based on 3-D data from a range camera.
an intelligent vehicle is proposed to reduce the number of To enable head position estimate and region-of-interest
fatalities in collisions caused by driver drowsiness. A recognition, the system integrates 2-D and 3-D approaches.
module for an advanced driver assistance system, which The points belonging to the heads are detected and extracted
caters to automatic driver drowsiness detection and driver for even further analysis based on the acquired cluster of 3-
distraction, is presented as a solution. To compute the D pixels from the sensor and examination of the 2-D
drowsiness and distraction indexes, artificial intelligence projection. The iterative closest points approach is then used
algorithms process visual information to locate, track, and to estimate head posture with a total of three freedom
analyse both the driver's face and eyes. As a result of a near- degrees (Euler angles). Eventually, significant facial areas
infrared lighting system, this real-time system operates are selected and used for additional analysis, such as action
during the night. Finally, examples of various driver images recognition and behavior analysis. The final application is a
captured in a real vehicle during the night are shown to low-cost 3-D driver surveillance system.
validate the proposed algorithms.
III. METHODOLOGY
In June 2012, A. Cheng et. al. described 'Driver
Drowsiness Recognition Based on Computer Vision The methodology for a driver drowsiness detection
Technology'. They demonstrated a non-invasive drowsiness and alert generation system can be divided into five stages.
detection method based on eye tracking and image
processing. A robust eye detection algorithm is introduced A. System Overview:
to address the issues caused by changes in brightness and The aim of a drowsiness detection system is to observe
driver posture addresses, and a sturdy eye detection a driver’s alertness level and alert them if they are becoming
algorithm is introduced. The percentage of eyelid closure, too drowsy so that they can take appropriate action to avoid
highest closure duration, blink frequency, the average a fatigue-related accident. If unfortunately accident is
opening level of the eyes, starting velocity, and closing occurred than the system also can alert emergency services
velocity of the eyes all are calculated. To reduce like ambulance if the driver got severe injuries. This can be
interrelations and retrieve an unbiased index, these measures done by sending a GPS location to the emergency services

IJISRT23MAY682 www.ijisrt.com 951

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
This is typically achieved by using a combination of Deep learning can be used to extract features from
cameras, sensors, image processing algorithms to detect audio, images and videos. It can be used to detect patterns
signs of drowsiness, such as yawning, head nodding, and and learn latent features that can be used to classify data and
eye closure, and API. The system is developed with help of make predictions. In audio, deep learning can be used to
machine learning algorithm like haar cascade algorithm. extract features like pitch, frequency, and rhythm. In images,
Haar cascade algorithms are a form of computer vision it can be used to detect edges, shapes, and textures. In the
widely used for detecting objects in an image or video. It is video, it can be used to identify objects, recognize faces, and
a machine-learning approach in which a cascade function is track motion. Additionally, deep learning algorithms can be
taught using a large number of both positive and negative used in feature extraction, such as dimensionality reduction,
images. Then it's used in other photos to detect movement. clustering and non-linear transformations. Using this
Using algorithm, we can easily detect drowsiness. technique ,we extract the particular part of driver face which
Developer API is also used to share the location of accident is used to find drowsiness .
are to the ambulance driver .
E. Modelling
B. Collecting Data: Once the features have been extracted, the next step is
The first step in developing a driver drowsiness to create a model that can accurately detect drowsiness. This
detection system is to collect data. This data should include can be done using machine learning algorithms such as
images or videos taken from a vehicle that captures the convolutional neural networks Convolutional neural
driver in different states of drowsiness including open eyes, networks (CNNs) are a kind of deep learning neural network
closed eyes, yawning, and nodding off. For collecting data that is used for image recognition and computer vision tasks.
we used observation technique which is used to collect They are based on the idea of how a human brain processes
moment of driver .we used two types of observation .there information by using a set of filters and feature detectors.
are; Using CNN, we predict drowsiness.
 Structured Observations: Structured observations involve
setting a predetermined set of behaviours to observe and IV. IMPLEMENTATION
record. For example, a researcher might set out to
observe and record how often a teacher smiles during a Driver drowsiness detection is a system that can detect
lesson. when a driver is falling asleep or is otherwise impaired due
 Unstructured Observations: Unstructured observations to fatigue or intoxication. This system uses various sensors,
involve observing and recording behaviours without any such as cameras, to monitor the driver’s facial expressions,
predetermined set of behaviours. For example, a eye movements, and other physiological signals. When the
researcher might observe and record all of the structure detects any variations in the driver’s behaviours
behaviours that a student displays while they are in a that indicate drowsiness, it can alert the driver to take a
classroom. break or pull off the road. The system proposed in the IEEE
paper utilizes a combination of a camera and a facial
C. Pre-processing Data: recognition algorithm. The camera captures images of the
After the data has been collected, it needs to be pre- driver’s face which are then analysed by the algorithm to
processed. This involves cleaning the data and removing any detect any variations in facial expressions and eye
noise from the images or videos. Additionally, the data movements that may indicate drowsiness. If a change is
should be labelled according to the driver's state of detected, the system can alert the driver to take a break or
drowsiness. We use some techniques to processing data. pull off the road. The system also utilizes several other
There are sensors, such as accelerometers and gyroscopes, to measure
 Data Cleaning: This involves identifying, correcting, and the driver's head movements. If the head movements
removing inaccurate or incomplete data. indicate that the driver is falling asleep, the system can alert
 Data Transformation: This involves transforming data the driver to take a break or pull off the road. The system
from one format to another in order to make it more also utilizes an EEG headset to measure the driver's
useful for analysis. brainwaves. This allows the system to detect changes in the
 Data Reduction: This involves reducing the size and driver's brainwaves that may indicate fatigue or intoxication.
complexity of the data in order to make it more If any changes are detected, the system can alert the driver
manageable and efficient to use. to take a break or pull off the road. The system also can alert
emergency services like ambulance if the driver got sever
By these techniques ,we built a complete dataset which injuries. This can be done by sending a GPS location to the
used in our system emergency services. We followed the following steps:
Step 1: Install and launch the OpenCV library.
D. Feature Extraction: Step 2: Capture the video frames from the camera and detect
After the data has been pre-processed, the next step is the face in each frame.
to extract features from the data. This involves extracting Step 3: Estimate the perspective ratio of the face also apply
important characteristics from the images and videos that the aspect ratio-based blinking algorithm.
can be used to identify drowsiness. Examples of features Step 4: If the duration of the blinks is less than a threshold
that can be extracted include eye opening, eye closing, and value, then the driver is considered to be sleepy and
yawning. For this we used deep learning technique. awareness should be sent to the driver.

IJISRT23MAY682 www.ijisrt.com 952

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Step 5: If the driver does not respond to the alert, then an we present a paradigm for locating driver fatigue in the
emergency alert should be sent to the specified contact list. context of communication-ready execution. The mechanism
Step 6: Finally, the system should be able to notice the is made up of three major sections: a camera, a facial
driver’s fatigue levels also alert the driver accordingly. recognition algorithm, and a message alert. The camera
Step7: If the driver got injuries due drowsiness, then with captures video frames of the driver's face and sends them to
changes in developer api like ambulance booking and the facial recognition algorithm. The facial recognition
tracking application, send location of that car to near by algorithm detects drowsiness in the driver by analyzing the
ambulance drivers to rescue the people . facial features of the driver. The message alert is triggered
when the facial recognition algorithm detects drowsiness in
V. FLOWCHART the driver, and it sends an aware message to the driver's
phone or other devices to alert them of the condition. In
The most important modeling tool is the data flow addition, the structure can be combined with other current
chart (DFC). It's used to simulate the system's components. vehicle safety systems, such as lane departure warning
The mechanism, the data used by the methodology, an systems, to further improve the safety of the driver. We
external entity that communicates with the system, and the evaluate our system with real-world data and compare it
data moves in the system are the components. with existing methods. The results show that our proposed
system can detect drowsiness in drivers with high accuracy.
VII. CONCLUSION

The use of drowsiness detection systems in vehicles is

an important development in the field of automotive safety,
as it can contribute to lowering the number of collisions
caused by drivers who become distracted or sleepy while
driving. In this paper, we present a summary of the existing
drowsiness detection systems and discuss their advantages
and limitations. We then propose an improved drowsiness
detection system that uses a mixture of many sensors and
machine learning algorithms to notice the onset of
drowsiness in drivers. Our proposed system is designed to
be robust and accurate, while also being affordable and easy
to use. Finally, we conclude our paper by discussing
potential future directions for this technology, including the
development of more advanced sensing and machine
learning algorithms that can detect more subtle signs of
drowsiness, as well as the use of this technology in the
development of autonomous vehicles.

REFERENCES

[1]. Powell N.B. and Schectman K.B. and Riley R.W, “The
Fig 1 Data Flow Chart (DFC)
road to danger: the comparative risks of driving while
sleepy,” The Laryngoscope, vol. 111, no. 5, pp. 887–
VI. RESULT AND DISCUSSION
893, May 2001.
[2]. World Health Organization, Global status report on
We are going to fit our camera into vehicles and, using
road safety: summary. World Health Organization,
software, we are going to detect drivers' facial reactions. If
2009. [Online]. Available
the driver is in sleepy mode, an alarm will blow to alert the
[3]. Ahmed, A. Pakistan Road Safety. Scribd. Retrieved
driver to wake up. It's more useful for drivers to escape from
November 4, 2021, from Pakistan-Road-Safety(2007,
accidents. It helps fully to save drivers' lives. Nowadays, we
June 21).
are facing many accidents to avoid implementing driver
[4]. Hong Suet. al. described ‘A Partial Least Squares
drowsiness detecting systems. The budget for our projects is
Regression-Based Fusion Model for Predicting the
less expensive. We are only going to spend money on
Trend in Drowsiness’,2008.
cameras. We are going to fit steering in vehicles to detect
[5]. Bin Yang et. al. ‘Camera based Drowsiness Reference
driver-face reactions. We are facing issues with the project,
for Driver State Classification under Real Driving
installing the project, and importing it. We can also do the
Conditions', June 2010.
project using IoT. It is more expensive and complex to
[6]. M.J. Flores et. al. ‘Driver drowsiness detection system
implement. Then we are fitting cameras into drivers' bodies
under infrared illumination for an intelligent
to detect the action of drivers. It affects the driver's body. It
vehicle,2011.
is also uncomfortable for drivers. Compared to Iot, the
[7]. A. Cheng et. al 'Driver Drowsiness Recognition
project is better to implement and has less cost. It is clear
Based on Computer Vision Technology', June 2012.
that although the use of this project driver drowsiness
detection system is to save the drivers' lives In this research,

IJISRT23MAY682 www.ijisrt.com 953

Volume 8, Issue 5, May – 2023 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
[8]. G. Kong et. al. ‘Visual Analysis of Eye State and Head
Pose for Driver Alertness Monitoring’, 2013.
[9]. Eyosiyaset. al.‘ Driver Drowsiness Detection through
HMM based Dynamic Modeling',2014, June.
[10]. Ji Q., Yang X.: Real-time eye, gaze, and face pose
tracking for monitoring driver vigilance. Real-Time
Imaging. 8(5):357-77 (2002).
[11]. Kyong Hee Lee, Whui Kim, Hyun Kyun Choi, Byung
Tae Jang, “A Study on Feature Extraction Methods
Used to Estimate a Driver’s Level of Drowsiness”,
International Conference on Advanced
Communications Technology (ICACT), 2019.
[12]. Ashish Kumar, Rusha Patra, “Driver Drowsiness
Monitoring System using Visual Behaviour and
Machine Learning”, IEEE Conference, 2018.
[13]. Cyun-Yi Lin, Paul Chang, Alan Wang, Chih-Peng Fan,
“Machine Learning and Gradient Statistics Based Real-
Time Driver Drowsiness Detection”, 2018 IEEE
International Conference on Consumer Electronics-
Taiwan (ICCE-TW).
[14]. Ching-Hua Weng, Ying-Hsiu Lai, Shang-Hong Lai,
“Driver Drowsiness Detection via a Hierarchical
Temporal Deep Belief Network”, In Asian Conference
on Computer Vision Workshop on Driver Drowsiness
Detection from Video, Taipei, Taiwan, Nov. 2016.
[15]. Anjith George, Aurobinda Routray, “Real-time Eye
Gaze Direction Classification Using Convolutional
Neural Network”, IEEE, 2016.
[16]. A. George and A. Routray, “Real-time Eye Gaze
Direction Classification Using Convolutional Neural
Network,” Int. Conf. Signal Process. Commun., pp. 1–
5, 2016
[17]. Venkata Rami Reddy Chirra, Srinivasulu Reddy
Uyyala, Venkata Krishna Kishore Kolli, “Deep CNN:
A Machine Learning Approach for Driver Drowsiness
Detection Based on Eye State”, Revue d'Intelligence
Artificielle Vol. 33, No. 6, December, 2019, pp. 461-
466.
[18]. Mkhuseli Ngxande, Jules-Raymond Tapamo, Michael
Burke, “Driver drowsiness detection using Behavioral
measures and machine learning techniques: A review
of state-of-art techniques”, 2017 Pattern Recognition
Association of South Africa and Robotics and
Mechatronics International Conference (PRASA-
RobMech).
[19]. T. Vogelpohla, M. Kühn, T. Hummel, M. Vollrath,
“Asleep at the automated wheel—Sleepiness and
fatigue during highly automated driving”,
https://doi.org/10.1016/j.aap.2018.03.013, Accident
Analysis and Prevention, 2018.
[20]. Xuan-Phung Huynh, “Detection of driver drowsiness
using 3D deep neural network and semi-supervised
gradient boosting machine”, in Asian Conference on
Computer Vision Workshop, 2016

IJISRT23MAY682 www.ijisrt.com 954