Driver Drowsiness Detection System Using Machine Learning

Abstract

We proposed to use this system to minimise the frequency of accidents caused by driver exhaustion, hence improving road safety. This device uses optical information and artificial intelligence to identify driver sleepiness automatically. We use Softmax to find, monitor, and analyse the driver\'s face and eyes in order to calculate PERCLOS (% of eye closure). It will also employ alcohol pulse detection to determine whether or not the person is normal. Due to extended driving durations and boredom in crowded settings, driver weariness is one of the leading causes of traffic accidents, particularly for drivers of big vehicles (such as buses and heavy trucks).

Introduction

I.  INTRODUCTION  

When a rider's ability to drive safely is harmed as a result of being physically or mentally fatigued or drowsy, this is known as driver fatigue. For the road transportation business, driver weariness is a serious safety threat. 

Too little sleep, driving when you should be sleeping, and working or being up for lengthy periods of time are the major reasons of "drowsy driving."

There are three types of methods for detecting driver drowsiness: 

1. Vehicle-based approaches,  
2. Behaviour-based approaches, and  
3. Physiological-signal based approaches.  

The physiological signals from a body, such as the electroencephalogram (EEG) for brain activity, the electrooculogram (EOG) for eye movement, and the electrocardiogram (ECG) for heart rate, are assessed to identify driver sleepiness in physiological techniques. Recent research have shown that approaches based on physiological signals (particularly the EEG signal) can identify driver tiredness with more reliability and accuracy than previous methods. In driving condition descriptions, the terms FATIGUE, drowsiness, and sleepiness are frequently interchanged. It's complex in nature, encompassing various human elements, and it's been tough for scholars to characterise throughout the years. Regardless of the uncertainty around weariness, it is an important element in driving safety. Fatigue has been identified as one of the top causes of road accidents throughout the world, according to research. It will also employ alcohol pulse detection to determine whether or not the individual is normal. It's especially important for professional drivers like bus and truck drivers. It is especially important for occupational drivers, such as bus and heavy truck drivers, who may be required to drive for lengthy periods of time during peak sleepiness hours. 

Bus drivers in the city encounter a stressful and taxing work environment on a regular basis, putting them at risk of driver fatigue. However, there has been a lack of research on the distinct origins and effects of this type of weariness. Much of the research into urban bus drivers has so far been done as part of major heavy vehicle driving studies, which involve a disproportionately large population of long-haul drivers who are expected to encounter a fundamentally distinct set of tiredness issues.  

A. OpenCv

OpenCV is an open source library used primarily for Computer Visibility Applications. This contains many functions and algorithms for tracking movements, facial recognition, object detection, classification and recognition and many other applications. Photos and streaming real-time videos can be customized to suit different needs using this library.

The library has more than 2500 advanced algorithms, which include your complete set of both old and modern computer views and machine learning methods. These algorithms can be used to detect and detect faces, to identify objects, to detect human actions in videos, to track camera movements, to track objects, to extract 3D objects, to produce 3D point clouds on stereo cameras, to combine images to produce high resolution. a picture of the whole scene, find the same images on the photo website, remove red eyes from photos taken using flash, follow eye movements, see the location and create tags to cover it with an abstract object, etc.

II. SYSTEM ARCHITECTURE

This paper presents a system to detect the driver’s drowsiness that works on grayscale images. The scheme of the system is presented in Fig. 5.1.

A. Modules  

1. Face Detection: For face recognition, we have used an algorithm that is part of the Computer Vision Toolbox System which is the Vision Cascade Detector. Which creates a system detector that detects an object using the Viola-Jones method. By default, the detector is configured to detect faces.

The steps to detect face area are as below:

a. Define and setup the cascade object detector using the constructor. The constructor uses built-in Viola-Jones algorithm to detect faces, noses, eyes, mouth and upper body.

b. Read the video or the image selected and run the face detector.

c. Draw the bounding box around the detected face.

2. Eye Detection: Because the drivers' faces may still be identified when they bend their heads, the eyes must be detected individually. The technique is the same as for face detection, except the object is changed to identify eyes. A difficulty arises when the author tries to use alternative films; other areas of the movie are mistakenly identified as eyeballs. True Positive is considered when the detected region is within the ocular area.

3. Mouth Detection

a. The purpose of characterizing the mouth is to detect signs of fatigue when yawning. The characteristics used to detect the mouth area are: Use the constructor to define and configure the cascading object detector. The constructor uses the built-in Viola-Jones algorithm to detect face, nose, eyes, mouth and upper body.

b. After watching a video or selecting a photo, launch the Face Recognizer.

c. Draw a bounding box around the identified mouth. The bounding box is the detection area.

III. RELATED WORK 

The advised machine is a driving force face tracking machine that detects driving force hypo vigilance through eye and face processing (each weariness and attention). Following the seize of a photograph, the primary degree of processing is face detection. Low vigilance indicators are then extracted from the face picture.

However, a specific eye detection section isn't used to decide the attention with inside the face; instead, a number of the maximum extensive signs associated with the attention area (top-1/2 of section of the face) are collected, making it computationally costly to rebuild the face popularity technique for all frames. It may also use alcohol pulse detection to peer if the man or woman is ordinary or not. Face monitoring algorithms are utilised to observe the driving force`s face in destiny frames till it's miles misplaced after the primary frame's face detection. 

IV.  METHODOLOGY

After surveying a number of different papers, the following methodologies have been identified:

A. Perclos

To identify the driver’s drowsy state using PERCLOS, we need to perform the following steps:

1. Perception of face and face pursuit.
2. Position of eye and eye pursuit.
3. Identification of the state of the eyes.
4. Calculation of percentage of eyelid closure.
5. Identification of the drowsy state.

PERCLOS is one of the measures to notice the state of drowsiness.

B. Viola Jones Algorithm

The Viola Jones algorithm uses the following techniques:

1. HAAR-based features
2. Integral Image Formation
3. AdaBoost technology
4. Class dividers

Features are selected based on pixel density in HAAR-based feature representation. It does not fit consideration, pixel values. HAAR is supported features are a scalar product between image and so on HAAR templates. Combined image format is used to calculate the feature. It looks at only four rooms in the picture. Adaptive boosting (AdaBoost) is used to select the required features. Due to the use of Adaptive Boosting there is a decrease algorithm calculation time. A cascade of classifiers is used to develop a strong chain of separation. The OpenCV Library offers a command a fast training tool called HAAR-training which produces a separator in XML format if provided good and bad examples of something to be found.

The following parameters are considered for performance analysis:

1. Drowsiness Detection Accuracy = total no. of times alert raised when eyes closed / (total no. of times alert raised when eyes closed + total no of times alert didn’t raise when the eyes are closed)
2. Yawn Detection Accuracy = total no. of times alarm raised when user yawns / (total no. of times alarm raised when user yawns + total no of times alarm didn’t raise when the user yawns).

VI.  FUTURE WORK  

Driver sleepiness is a major concern in today's culture, as the drowning problem is causing an increase in accidents on a regular basis. With the help of Neural Networks and other real-time sensor devices, it will be deployed in the future. So that more accuracy may be achieved. 

1. School bus drivers found the strategy to be incredibly useful. 
2. It will also use alcohol pulse detection to assess whether the person is normal. 

Conclusion

The rising incidence of traffic accidents caused by less driver alertness has become a severe societal issue. According to statistics, 20 percent of all road accidents are caused by drivers who are not paying attention. Furthermore, incidents involving driver hypo-vigilance are more dangerous than other types of collisions, since drowsy drivers frequently fail to take the necessary precautions before a collision. As a result, establishing methods for monitoring driver attentiveness and warning the driver when he is tired and not paying attention to the road is critical to avoiding accidents. It will also employ alcohol pulse detection to determine whether or not the person is normal. In the subject of active safety research, preventing such mishaps is a key focus of work. Changes in facial characteristics such as the eyes, head, lips, and face are visible in those who are tired. To check a driver\'s alertness, computer vision can be a natural and nonintrusive tool. Faces, being the fundamental form of human communication, have long been a focus of computer vision research. One of the most promising commercial applications of automated facial expression recognition is the identification of driver weariness. Face detection, facial expression information extraction, and expression categorization are the three tiers of tasks involved in automatic facial expression identification (or analysis). The key difficulty in these jobs is information extraction for feature-based face expression identification from a picture sequence. It entails detecting, identifying, and tracking facial feature points in a variety of lighting conditions, face orientations, and expressions. In this study, an SVM Classifier is used to identify weariness and provide various outcomes. The work\'s accuracy is 70 percent in this case.

References

[1] M.Omidyeganeh , A.Javadtalab,S . Shirmohammadi , “Intelligent Driver Drowsiness Detection through Fusion of Yawning and Eye Closure”,IEEE International Conference on Virtual Environmnets , Human-Computer Interfaces, And Measurements Systems (2011) [2] Menchie Miranda , Alonica Villanueva , Mark Jomar Buo , Reynald Merabite , Sergio Paulo Perez , John Michael Rodriguez , “Portable Prevention and Monitoring of Driver’s Drowsiness Focuses to Eyelid Movement Using Internet of Things”,IEEE International Conference on Humanoid, Nanotechnology, Information Technology, Communication and Control, Environment, and Management (2018) [3] Yashika Katyall , Suhas Alur , Shipra Dwivedi , “Safe Driving By Detecting Lane Discipline and Driver Drowsiness”, IEEE International Conference on Advanced Communications, Control and Computing Technologies (2014) [4] Bappaditya Mandal , Liyuan Li , Gang Sam Wang and Jie Lin, “Towards Detection of Bus Driver Fatigue Based on Robust Visual Analysis of Eye State, IEEE Transactions on Intelligent Transportation Systems(2016) [5] Vidhu Valsan A , Paul P Mathai , Lerin Babu “Monitering Driver’s Drowsiness Status at Night Based on Computer Vision” International Conference on Computing, Communication, and Intelligent Systems ICCCIS(2021)

Copyright

Copyright © 2022 Ms. K. G. Walke, Harshvardhan Shete, Amjad Mulani, Darshan Asknani, Rushikesh Gadekar. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.