tracking of driver drowsiness detection and alert system ...journalstd.com/gallery/3-may2020.pdf ·...

Tracking of driver drowsiness detection and alert system using

Machine Learning

MUNDURU SYAMALAKUMARI1

Assistant Professor

Department of Computer Science and Engineering

Sri Sivani Institute of Technology, Srikakulam, Andhra Pradesh, India.

Email – [email protected]

SAHUKARA GAURAV2, VANA HIMABINDU

3, KARRI NEELAVENI

4, BAGGU MOUNIKA

5

Department of Computer Science and Engineering

Sri Sivani Institute of Technology, Srikakulam, Andhra Pradesh, India.

Abstract:

More and more occupations today need commitment over the long term. Drivers must keep a close eye on

the lane, so they can quickly respond to unexpected events. Driver exhaustion is also the primary cause of

frequent traffic incidents. There is also a need to establish systems that can identify and warn a driver of

her / him bad psychophysical state, which could dramatically reduce the number of fatigue-related car

accidents. Developing these systems, however, encounters several difficulties related to swift and proper

identification of the fatigue symptoms of a driver. Usage of the vision-based approach is one of the

technological possibilities to incorporate driver drowsiness monitoring systems. The article describes the

drowsiness identification devices currently in use for drivers. Here we detect the driver's sleepiness by

estimating his vision system. And if the driver is sleepless a warning message to send to the registered

user and a latitude and longitude positions send to the police control room to identify the location of

driver to avoid the problem

Keywords: Drowsiness detection, Eye blink , Google map, face detection

1. Introduction

Driver depletion is a huge variable in a broad number of vehicle mishaps. Late experiences, survey

that yearly 1,200 passing's and 76,000 wounds can be credited to exhaustion related mishaps. Street

Accidents in Sri Lanka cause of money related misfortunes worth around Rs.9.34 billion consistently. It

very well may be seen there are around 2,400 street mishaps reliably which is one demise for each at regular

intervals. It has been figured around 20% of vehicle crashes with driver fatalities are because of driver's

sluggishness. It was revealed that driving execution rapidly drop with extended tiredness which bring about

creation over 20% of all vehicle mishaps. Less consideration heads the driver to being occupied and the

probability of road mishap goes high.

Drowsiness-related crashes have all the earmarks of becoming more extreme due to the fatigue

involved in the higher speeds and the driver was unable to take any avoidance operation or even stop until

the crash. Improving technology to identify or avoid driver fatigue is an effective check in the area of

accident-prevention systems. Because of the danger posed by that drowsiness on the lane, strategies need to

be developed to test its influences. Loss of consciousness due to the tiredness induces certain shifts in the

body and behaviors of the human being.

Science, Technology and Development

Volume IX Issue V MAY 2020

ISSN : 0950-0707

Page No : 14

mailto:[email protected]

Such side effects and parameters allow us to assess the degree of drowsiness effectively. Specific

methods for recognizing somnolence can be grouped into two general classifications. In the first meeting,

the strategies understand the level of tiredness based on the physiological changes in the body. In the first

grouping, eye state, expression properties, time interval between two yawning, head position, sitting

carriage, heart rate and brain signals are only a few examples of the techniques. In addition, Drowsiness

triggers some driving style changes. Second group techniques measure the driver's level of drowsiness by

following those progressions. When part of the second classification strategy, steering angle, distance from

the following vehicle, vehicle lateral direction, longitudinal speed, longitudinal speed up, and lane

departure are used.

A new approach to automotive safety and protection is being projected during this design with

autonomous area mainly based automatic automotive system. A Drowsy Driver Detection System and a

traffic detection system with an interference of external vehicles dodging mainly architecture dependent.

Related car fatigue accidents have quite increased in recent times. In order to alleviate these issues, we

implemented the driver alert system by still watching through driver's eyes as sensing as the driver's state

of affairs is predicted based primarily on the native recognition-based AI system. For a significant number

of car collisions driver fatigue is an important factor. Recent statistics estimate that annually1,200 deaths

and 76,000 injuries can be attributed of fatigue related crashes.

2. Literature Survey

Driver drowsiness is a major problem among commercial truck drivers and is responsible for

thousands of injuries and fatalities each year. The National Highway Traffic Safety Administration

(NHTSA)'s Office of Crash Avoidance Research (OCAR) described driver drowsiness as one of the

leading causes of single- and multiple-car accidents in a 1994 study (Knipling 1994).NHTSA reports that

100,000 collisions annually include driver fatigue resulting in more than 40,000 accidents. Two types of

conditions can be used to assess driver's drowsiness. Variables of physical and physiological condition and

of the engine. Physical and physiological measurements include brain wave or Electroencephalogram

(EEG) measurements (Akerstedt and Gillberg 1990). Skipper, Wierwille et al. 1984; Dingus, Hardee et al.

1985; Ueno, Kaneda et al. 1994; Ogawa and Shimotani 1997). PERCLOS (PERcent eyelid CLOSure) is

one of the most commonly recognized tests for the assessment and identification of drowsiness in medical

literature (Dinges, Mallis et al. 1998; Grace, Byrne et al. 1998).

Drowsiness detection systems have been developed that operate on the basis of physical and physiological

characteristics measurement and can provide very good accuracy in detection. We may have some

drawbacks, however. The problem with an EEG is that it needs electrodes to be connected to the scalp,

making the usage very impractical. Fatigue was estimated at 15 per cent of single-vehicle fatal truck

accidents (Wang and Knipling 1994) and is the most common cause to accidents causing fatal injury to a

truck driver (NTSB 1990). Based on data from the NHTSA General Estimates System (GES) (Knipling

and Wierwille 1994), while the number of drowsiness-related collisions involving passenger vehicles

exceeds that of combination-unit trucks, The number of vehicle life cycle involvements for trucks is around

four times higher due to their very high degree of visibility, as well as the increased probability of driving

at night. As regards the measurement of Vehicle State Variables, other methods to detect driver drowsiness

are based on tracking driver inputs or variables of vehicle performance while driving. Such approaches



ISSN : 0950-0707

Page No : 15

have the advantage of making drivers non-intrusive. Wierwille et al. (1992) Robert Gabriel Lupu has

addressed that several eye pupil / iris detection algorithms were created in the previous year. There are two

methods, namely based on ambient or infrared light, depending on the source light point of view.

Dongheng Li, Derick J. Parkhrust discussed the Starburst algorithm is a robust eye-tracking

algorithm the blends feature-based and model-based approaches to achieve a reasonable trade-off between

runtime efficiency and accuracy for dark-pupil infrared imagery. V. Starovoitov and D. Geometric learning

methods have been explored by Samal extracting distinctive geometric features from images. These

features may be corner features, edge features, blobs, ridges, image texture at salient points, and so on,

which can be identified using feature detection methods.

2.1 Techniques

2.2.1 Techniques for Detecting Drowsy Drivers

Possible techniques for detecting drowsiness in drivers can be generally divided into the

following categories: sensing of physiological characteristics, sensing of driver operation, sensing of

vehicle response, monitoring the response of driver.

2.2.2 Monitoring Physiological Characteristics

Among these methods, the techniques that are best, based on accuracy are the ones based on human

physiological phenomena. This technique is implemented in two ways: measuring changes in physiological

signals, such as brain waves, heart rate, and eye blinking; and measuring physical changes such as sagging

posture, leaning of the driver’s head and the open/closed states of the eyes.

2.2.3 Other Methods

Driver operation and vehicle behavior can be implemented by monitoring the steering wheel movement,

accelerator or brake patterns, vehicle speed, lateral acceleration, and lateral displacement. The set are non-in

trust of detecting drowsiness, but are limited to vehicle type and driver conditions

2.3 Methods Focusing on Driver's Performance

In order to detect drowsiness, studies on driver’s performance use lane tracking, distance between driver’s

vehicle and the vehicle in front of it; place sensors on components of the vehicle such as steering wheel, gas

pedal and analyze the data taken by these sensors.

2.4 Methods Focusing on Driver's State

The methods use physiological signals such as Electroencephalography (EEG), heart rate variability (HRV),

pulse rate and breathing. The spectral analysis of heart rate variability shows that HRV has three frequency

bands: high frequency band (0.15- 0.4 Hz), low frequency band (0.04-0.15 Hz) and very low frequency

band (0.0033- 0.04Hz).

2.5 Methods Using Computer vision

This group of methods are not offensive and does not make any disturbance to the driver, that’s why

these methods are more preferable. These methods are separated into two groups: the ones using in fared



ISSN : 0950-0707

Page No : 16

illumination and the ones using day illumination. The former finds the location of the eyes and detect eye

states by making use of retinal reflections of infrared waves. Matsuo and Khiat, who work for NISSAN,

divided driver’s condition into 4 categories: normal (alert), slightly sleepy, intensely sleepy and drowsy.

Fig 1: Flowchart of eye state recognition

2.6 Approach to The Solution

The method we propose belongs to the group which focuses on driver’s state by making use of computer

vision. Eye closure rate, in other words, percentage of eye closure (PERCLOS) is a reliable measure to

detect drowsiness [48]. This thesis makes use of PERCLOS to decide whether the driver in a video segment

is drowsy or alert. For every frame in the video segment, eye state estimation is performed. There are 3

states of an eye: open eye, semi-closed eye and closed eye. The estimations for each frame in a video

segment are combined and the driver’s state is estimated.

The video segment is extracted to its frames. After extraction of the frames of the video segment, the

frames are input to the part called eye region extractor. Eye region extractor firstly finds the candidates for

right and left eye regions and face by making use of extended version of Viola-Jones algorithm, which is

available in Computer Vision System Toolbox of Python. Among the candidates of face, the wrong

candidates are eliminated by some decision rules and the face region is detected. Combining the estimations

for right and left eyes increases the accuracies for both eye state and drowsiness detection. Since

combination of the estimations for right and left eyes is not a common method used in the literature,

increasing the accuracy with this method is a contribution of our proposed algorithm. Most of the studies

assign eyes only two states: open and closed. As another contribution, this study reveals the fact that semi-



ISSN : 0950-0707

Page No : 17

closed state has an important role in detecting drowsiness and defining three states instead of two states

increases the accuracy of the drowsiness detection method proposed.

3. Material and Method

In this study, real-time camera image data of four volunteers were used to test the driver’s

sleepiness. Camera images are taken from a driving simulation. The driver images taken were reduced

to 320x240 spatial resolution and converted to gray scale with th In this Equation, Y represents pixel

brightness, R is red, G is green and B is blue color brightness.

Since color information is not used in the image processing algorithms to be used, only the gray

scale image is processed. The face and eye images of the driver are detected and cropped by the Viola-

Jones detector method. The Viola-Jones detector is an AdaBoost classifier that uses Haar-like features.

AdaBoost classifiers train T amount of ht weak classifiers which usually consist of independent and

single level binary decision trees.

A not weight value is given for each classifier. As the input dataset if feature vectors labeled

with a binary tag are used, each of which is only -1 and +1. Finally, the class of the xi input is

calculated by Equation

In this equation, H(x) is the class of the x sample, ht is the weak classifier, and αt is the classifier

coefficient. The Sign function returns 1 for all positive values and -1 for all negative values. Zero values

returned as zero. An example showing the application of Haar-like features to the facial image is shown

in Figure1.After finding the facial image, the left and right eye areas were cropped using the geometric

ratios of the face and the left and right eye images were simultaneously found in these areas using

parallel tasks. The eye images were combined to obtain a minimum size image containing both eyes.

40 Gabor filters were used at eight different angles and five different scales to extract the

features of these images. Gabor filters define sine and cosine functions within a Gaussian window. Two

dimensional Gabor wavelets are obtained by using two dimensional forms of these functions. The real

(even) and imaginary (odd) components of two- dimensional Gabor wavelets are denoted as Applying

Haar-like feature to recognize facial structure.



ISSN : 0950-0707

Page No : 18

The total duration of the spontaneous closing and opening phases of the human eyelid is approximately

334 milliseconds The classification speed of the IBk algorithm is almost half the speed of the blink time.

However, the J48 decision tree has more than enough speed to catch an eye blink easily. For this reason,

we concluded that the J48 algorithm, which is 2.78% lower in accuracy, is suitable for real- time

classification. A real-time software has been developed for detection of driver drowsiness using personal

data. This software classifies the eyes as open or closed by cropping the driver’s face and eye images in

a real-time video image through the model trained beforehand by machine learning algorithms.

Fig 2: example haar cascade classifier Fig 3: pattern cascade classifier

Recognizing as classifier

The procedure is :

1. Eye localization.

2. Thresholding to find the whites of the eyes.

3. Determining if the “white” region of the eyes disappears for a period of time (indicating

blink).

4. Eye blink detection with OpenCV, Python, and dlib

5. Our blink detection blog post is divided into four parts.

6. In the first part we’ll discuss the eye aspect ratio and how it can be used to determine ifa

person is blinking or not in a given video frame.

7. From there, we’ll write Python, OpenCV, and dlib code to perform facial landmark

detection and detect blinks in video streams.

8. Based on this implementation we’ll apply our method to detecting blinks in example

webcam streams along with video files.



ISSN : 0950-0707

Page No : 19

4. Understanding The “Eye Aspect Ratio” (EAR):

Fig 4: calculating eye aspect ratio

Fig 5: graph representation of eye aspect ratio

We can apply facial landmark detection to localize important regions of the face, including eyes,

eye brows, In terms of blink detection, we are only interested in two sets of facial structures — the

eyes. Each eye is represented by 6(x,y)-coordinates, starting at the left-corner of the eye(as if you were

looking at the person), and then working clockwise around the remainder of the region: In this I

demonstrated how to build a blink detector using OpenCV, Python, and dlib.) The first step in building

a blink detector is to perform facial landmark detection to localize the eyes in a given frame from a

video stream. Once we have the facial landmarks for both eyes, we compute the eye aspect ratio for

each eye, which gives us a singular value, relating the distances between the vertical eye landmark

points to the distances between the horizontal landmark points. Once we have the eye aspect ratio, we

can threshold it to determine if a person is blinking — the eye aspect ratio will remain approximately

constant when the eyes are open and then will rapidly approach zero during a blink, then increase again

as the eye opens. To make our blink detect or more robust to these Computing the eye as pectratio for

the N-thframe, along with the eye aspect ratios for N – 6 and N + 6 frames, then concatenating these

eye aspect ratios to form a 13 dimensional feature vector.

Where p1, …, p6 are 2D facial landmark locations. The numerator of this equation computes the

distance between the vertical eye landmarks while the denominator computes the distance between

horizontal eye landmarks, weighting the denominator appropriately since there is only one set of

horizontal points but two sets of vertical points.



ISSN : 0950-0707

Page No : 20

https://pyimagesearch.com/2017/04/10/detect-eyes-nose-lips-jaw-dlib-opencv-python/

5. Algorithm Stages Image Capture

Utilizing a web camera introduced inside the automobile we can get the picture of the driver.

Despite the fact that the camera creates a video clip, we have to apply the developed algorithm on each

edge of the video stream. This paper is only focused on the applying the proposed mechanism only on

single frame. The used camera is a low cost web camera with a frame rate of 30 fps in VGA mode.

Logitech Camera is used for this process is shown in figure.

Fig 6: Camera used for implementing drowsiness detection system

Dividing into Frames

We are dealing with real time situation where video is recorded and has to be processed.

But the processing or application of algorithm can be done only on an image. Hence the captured

video has to be divided into frames for analyzing.

Object Detection

Object detection is commonly defined as method for discovering and identifying the existence

of objects of a certain class. Also it can be considered as a method in image processing to find out an

object from images. There are several ways to classify and find objects in a frame. Out of that one way

can be based on color identification. But it i snot an efficient method to detect the object as several

different size object of same color may be present. Hence a more efficient way is Haar-like features,

developed by Viola and Jones on the basis of the proposal by Papageorgiou et. al in 1998.The output of

each stage is labeled as either positive or negative–positive meaning that an object was found and

negative means that the specified object was not found in the image.

Face Detection

In this stage we detect the region containing the face of the driver. A specified algorithm is for

detection of eyes in every frame. By face detection we mean that locating the face in a frame or in other

words finding location of facial characters through a type of technology with the use of computer. The

frame may be any random frame. Only facial related structures or features are detected and all others

types of objects like buildings, tree, bodies are ignored.



ISSN : 0950-0707

Page No : 21

Eye Detection

After successful detection of face eye needs to be detected for further processing. In our

method eye is the decision parameter for finding the state of driver. Though detection of eye may

be easier to locate, but it’s really quite complicated.

At this point it performs the detection of eye in the required particular region with the use of

detection of several features. Generally, Eigen approach is used for this process. It is a time taking

process. When eye detection is done then the result is matched with the reference or threshold value for

deciding the state of the driver.

6. Experimental Protocol

The system proposed is built on Linux Operating system and the detection mechanism is

carried out with the help of OpenCV Library. The different phases of the algorithm are driven by:

Face Detection

Eye Detection

Eye Closure Characterization

Yawn Detection

7. Analysis

Analysis of the Existing system, helps in designing problem statement of Proposed

system. In the following section based on the analysis of existing system, the requirements of

proposed system have been defined.

7.1 Existing System

SVM (support vector machine) was used to classify the components in the input video.

While cropping the region of interest components in the video is not accurate. Sometimes it will

show regions wrong. To sense the eyes first we have to create boundary boxes for that and a

classification algorithm. The algorithm of SVM will not support.

By using a non intrusive machine vision based concepts, drowsiness of the driver

detected system is developed. Many existing systems require a camera which is installed in front

of driver. It points straight towards the face of the driver and monitors the driver's eyes in order

to identify the drowsiness. For large vehicle such as heavy trucks and buses this arrangement is

not pertinent. Bus has a large front glass window to have a broad view for safe driving. If we

place a camera on the window of front glass, the camera blocks the frontal view of driver so it is

not practical. If the camera is placed on the frame which is just about the window, then the

camera is unable to detain the anterior view of the face of the driver correctly.



ISSN : 0950-0707

Page No : 22

The open CV detector detects only 40% of face of driver in normal driving position in

video recording of 10minutes.In the oblique view, the OpenCV eye detector(CV-ED) frequently

fails to trace the pair of eyes. If the eyes are closed for five successive frames the system

concludes that the driver is declining slumbering and issues a warning signal. Hence existing

system is not applicable for large vehicles. In order to conquer the problem of existing system,

new detection system is developed in this project work.

7.2 Proposed Methodology

There are several different algorithms and methods for eye tracking, and monitoring.

Most of them in some way relate to features of the eye (typically reflections from the eye) within

a video image of the driver. The original aim of this project was to use the retinal reflection as a

means to finding the eyes on the face, and then using the absence of this reflection as a way of

detecting when the eyes are closed. Applying this algorithm on consecutive video frames mayaid

in the calculation of eye closure period. Eye closure period for drowsy drivers are longer than

normal blinking. It is also very little longer time could result in severe crash. So we will warn the

driver as soon as closed eye isdetected.

Drowsiness detection approach

Fig 7: Flow chart showing entire process of drowsiness detection system

Alarm

Eye

detection



ISSN : 0950-0707

Page No : 23

Fig 8: capturing the eyes Fig 9: counting the blinking rate

Fig 10: alert send to mail registered Fig 11:tracking the location using maps

8. Conclusion

The driver abnormality monitoring system developed is capable of detecting drowsiness,

drunken and reckless behaviors of driver in a short time. The Drowsiness Detection System

developed based on eye closure of the driver can differentiate normal eye blink and drowsiness and

detect the drowsiness while driving. The proposed system can prevent the accidents due to the

sleepiness while driving. The system works well even in case of drivers wearing spectacles and even

under low light conditions if the camera delivers better output. Information about the head and eyes

position is obtained through various self-developed image processing algorithms. During the

monitoring, the system is able to decide if the eyes are opened or closed. When the eyes have been

closed for too long, a warning signal is issued. processing judges the driver’s alertness level on the

basis of continuous eye closures.



ISSN : 0950-0707

Page No : 24

9. Further Enhancement

In this work, for both of the features sets, we used simple mean, min, max, and standard

deviations for our aggregation functions. It remains future work to try other aggregate functions such

as Fourier transforms and wavelets to see if they can improve classification. For the ground truth, we

assumed that if the driving is done at late night, then the driver is drowsy. This is a strong

assumption and may not be the case in the real world. A driver may be drowsy at parts of the run and

awake on other parts of the run. Therefore, for future work, we would like to try and detect parts of

the run where the driver is drowsy. This is a more challenging task and requires more complicated

features.

10. References

[1] Miaou, “Study of Vehicle Scrap page Rates,” Oak Ridge National Laboratory, Oak Ridge, TN,,

S.P.,April 2012.

[2] Wreggit, S. S., Kim, C. L., and Wierwille, W. W., Fourth Semi-Annual Research Report”, Research

on Vehicle-Based Driver Status Performance Monitoring”, Blacksburg, VA: Virginia Polytechnic

Institute and State University, ISE Department, January 2013.

[3] Bill Fleming, “New Automotive Electronics Technologies”, International Conference on Pattern

Recognition, pp. 484- 488,December 2012.

[4] Ann Williamson and Tim Chamberlain,“Review of on-road driver fatigue monitoring devices”, NSW

Injury Risk Management Research Centre, University of New South Wales, , July 2013.

[5] E. Rogado, J.L. García, R. Barea, L.M. Bergasa, Member IEEE and E. López, February, 2013, “Driver

Fatigue Detection System”, Proceedings of the IEEE International Conference on Robotics and

Biometics, Bangkok, Thailand.

[6] Boon-Giin Lee and Wan-Young Chung, Member IEEE, “Driver Alertness Monitoring Using Fusion

of Facial Features and Bio-Signals”, IEEE Sensors Journal, VOL. 12, NO. 7, July 2012.

[7] H. Singh, J. S. Bhatia, and J. Kaur, “Eye tracking based driver fatigue monitoring and warning

system”, in Proc. IEEE IICPE, New Delhi, India, Jan. 2014.

[8] M. Hemamalini, P. Muhilan“Accident preventionusing eye blink sensor”, vol 1, Issue L11, 2017.

[9] RamalathaMarimuthu, A. Suresh, M. Alamelu and S.Kanagaraj “Driver fatigue detection using

image processing and accident prevention”, International journal of pure and applied mathematics, vol.

116, 2017.

[10] TejaswiniJagdale, PradnyaJadhav, PrajaktaTotre,MayuraZadane, ShrilekhaMankhai “Driver

drowsiness detection, alcohol detection and accident prevention”, IJET, vol3 issue1, jan 2017

[11] BappadityaMandal, LiyuanLiyuan Li, Gang Sam Wang and JieLin ”Towards detection of bus

driver fatigue based on robust visual analysis of eye state”,IEEE transaction on intelligent transportation

systems,2016.

[12] Suhaskatkar, Mahesh ManikKumbhar, PritiNavanathKadam”Accident prevention system using eye

blink sensor”, IRJET, Vol.03 Issue 05, 2016.



ISSN : 0950-0707

Page No : 25

[13] Tejasweenimusale, prof B,H. Pansambal, ”Real time driver drowsiness detection system using

image processing”, IJREAM, Vol 02, Issue 08, 2016.

[14] Omkar, RevatiBhor, PranjalMahajan, H.V. Kumbhar “Survey on Driver‟s drowsiness detection

system”,vol.132,2015.

[15] Christy, Jasmeen Gill, “A Review: Driver drowsiness detection system”, IJCST, Vol.3 Issue 4,jul-

aug 2015.

11. About Authors

1. MUNDURU SYAMALAKUMARI, M.TECH. Working as an Assistant Professor of Computer Science

and Engineering Department in Sri Sivani Institute of Technology. She is having 5 years of teaching

experience. Her areas of interest are Operating Systems, Data Mining, Artificial Intelligence, Data

Structures, Computer Networks, Compute Graphics, etc.,

Email- [email protected]

2. SAHUKARA GAURAV, B.TECH. Student of Sri Sivani Institute of Technology in Computer Science

and Engineering Department. He has interest in learning modern technologies. He acted as a student

coordinator in organizing several seminars, Paper presentations and Symposiums at college. He participated

many workshops conducted in several Engineering colleges. His areas of interest are Datamining, Big Data,

AI, Machine Learning, Key Generation Algorithm, Computer Networks, Cloud Computing and Image

Processing.


3.VANA HIMABINDU, B.TECH. Student of Sri Sivani Institute of Technology in Computer Science and

Engineering Department. She has interest in learning modern technologies. Her areas of interest are

Machine Learning, Data Mining, Big Data and Computer Architecture. Email- [email protected]

4.KARRI NEELAVENI, B.TECH. Student of Sri Sivani Institute of Technology in Computer Science and


Machine Learning, Data Mining and Information Security. Email- [email protected]

5.BAGGU MOUNIKA, B.TECH. Student of Sri Sivani Institute of Technology in Computer Science and


Machine Learning, Image Processing and Data Mining.




ISSN : 0950-0707

Page No : 26






tracking of driver drowsiness detection and alert system ...journalstd.com/gallery/3-may2020.pdf ·...

Documents