Driver Drowsiness Detection

Download Full-Text PDF Cite this Publication

Text Only Version

Driver Drowsiness Detection

V B Navya Kiran

Dept. of Information Science and Engineering JSS Academy of Technical Education Bengaluru, India

Anisoor Rahman

Dept. of Information Science and Engineering JSS Academy of Technical Education Bengaluru, India

Raksha R

Dept. of Information Science and Engineering JSS Academy of Technical Education Bengaluru, India

Varsha K N

Dept. of Information Science and Engineering JSS Academy of Technical Education Bengaluru, India

Dr Nagamani NP

Assistant Professor

Dept. of Information Science and Engineering JSS Academy of Technical Education Bengaluru, India

Abstract – Machine learning techniques have been used in order to predict the condition and emotion of a driver to provide information that will improve safety on the road. It is an application of artificial intelligence. Artificial Intelligence is a method by which systems can automatically learn as well as improve without being explicitly programmed. A drivers condition can be estimated by bio-indicators, behavior while driving as well as the expressions on the face of a driver. In this paper we present an all-inclusive survey of recent works related to driver drowsiness detection and alert system. We also present the various machine learning techniques such as PERCLOS algorithm, HAAR based cascade classifier, OpenCV which are used in order to determine the drivers condition. Finally, we identify the challenges faced by the current systems and present the corresponding research opportunities.

Keywords – Artificial Intelligence, Autonomous Vehicle Technology, Drowsiness Detection, Machine Learning.

  1. INTRODUCTION

    Sleep deprived driving is the operation of a motor vehicle while being psychologically weakened due to a lack of sleep. Sleep deprived driving is a major cause of vehicular accidents. When a person does not get the required amount of sleep, their ability to properly function is affected. When their ability to function is affected, they have longer reaction time and their memory and judgement is weakened. Many studies have found that sleep deficiency can affect driving as much as alcohol inebriation. About 20% of the people have admitted to falling asleep at the wheel with 40% of the people confessing that this has taken place at least once in their driving careers. Research shows, in India, 40% of highway crashes or near crashes occur due to drowsy driving whereas more than 50% of all deadly highway crashes which involve more than two cars are alcohol related. More than 65% of all deadly single car crashes are related to inebriation. Looking at these statistics, it is imperative that we develop a driver

    safety system. In order to develop such a system, we need to estimate the condition of the driver at wheel. The following is a concise description of the papers surveyed.

    The following is a concise description of the papers surveyed. The paper presents an arithmetic based method to solve the problem related to the detection of drowsiness. Three stages were involved. They are Face detection, Eye position detection and Eye tracking. This paper provides an efficient method for the detection of the state of the driver [2]. This framework uses the motion of the eyes to detect the state of the driver and gives an alert within 0.5 seconds. The performance of the driver is transcribed in the form of a graph [6]. A new method for fatigue detection is presented. YCbCr colour space and canny edge detection methods are used. These methods are used to determine if the driver is under fatigue. When the driver is drowsy, an alarm system is turned on [9]. A distinct system which focuses on the concept of computer vision is designed. A software algorithm is developed. This algorithm is partially tested and is found to be working effectively. Research is in progress in order to develop a full-blown system [4]. The developed system is capable in identifying the state of drowsiness quickly. The system is capable of differentiating between normal eye blink and the eye blink associated with drowsiness. It is capable of performing under low light conditions and when the driver is wearing spectacles. This can further be developed by adding different sensors [5]. The developed system is based on computer vision. The system utilizes Viola Jones algorithm as well as the CAMSHIFT algorithm [8]. This paper is concerned with the development of a software framework for the timely and precise detection of drowsiness. Multiple facial features were considered as inputs [3]. The paper proposes a method for detecting the drowsy state based on the time-series analysis of the angular velocity of the steering wheel. When compared with the traditional method, this method provides a series of advantages [10].It groups drowsiness detection techniques into two kinds. They are

    driver based and vehicle based. It also provides a survey of numerous driver and vehicle-based techniques [11]. A drowsiness detection system which is dependent upon an algorithm known as shape predictor algorithm and eye blink rate is developed. It is based on the concept of image processing. The system provides a non-invasive approach. This system also proposes the incorporation of yawning as a parameter to detect drowsiness [7]. In order to estimate a drivers condition, certain facial features were identified. Using python libraries they were examined. These features were rate of eye closure, ECD, per closure, head positions and rate of yawning. Certain limitations were also highlighted [1].

  2. LITERATURE REVIEW

    The developed system is a real time system. It uses image processing for eye and face detection. HAAR based cascade classifier is used for face detection. An algorithm to track objects is used to track the eyes continuously. In order to identify the drowsy state of the driver, the PERCLOS algorithm issued [2]. The paper focuses on developing a non- intrusive system which can detect fatigue and issue a warning on time. The system will monitor the drivers eyes using a camera. By developing an algorithm, the symptoms of driver fatigue can be detected early enough to avoid accident. When the signs of fatigue have been identified output in the form of sound and seat belt vibration is provided to alert the driver. Warning will be deactivated manually rather than automatically. This paper uses a faster algorithm than PERCLOS. This system will detect drivers fatigue by the processing of the eye region. After image acquisition, the first stage of processing is face detection. If eyes are blinking

    visual, non-visual, and vehicular features into one. The last idea is to develop wearable hardware such as smart watches in order to detect drowsiness [11].

  3. METHODOLOGY

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

    1. PERCLOS

      Initially, in order to identify the drivers drowsy state using PERCLOS, we need to perform the following steps as per [2]:

      • Perception of face and face pursuit.

      • Position of eye and eye pursuit.

      • Identification of the state of the eyes.

      • Calculation of percentage of eyelid closure.

      • Identification of the drowsy state.

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

    2. CAMSHIFT

      A robust and nonparametric technique is used [2]. It implements the CAMSHIFT algorithm. CAMSHIFT (continuously adaptive mean-shift) is an efficient and light- weight tracking algorithm. It is based on the concept of mean shift It is suitable for trackin targets in simple cases. It is not efficient in tracking objects in complex situations. A detection algorithm can be applied to successive frames of a video sequence to track a single target [2]. According to [2] the detection algorithm can be described by the following steps:

      1. Initialize the size as well as the position of the search window.

      2. Calculate the mass centre (X , Y ) of the window.

        normally no warning is issued. If the eyes are closed for more c c

        than 0.5 seconds, this system issues warning to the driver. The warning is in form of an alarm and vibration. MATLAB is used for the processing of the image [6]. System makes use of the number of eye blinks for detecting the state of drowsiness in a driver. The system makes use of OpenCV and Raspberry Pi module with a single camera view. The eye status is obtained through image processing algorithms. This paper takes into account only the state of the eyes, it does not focus on the frequency of yawning [7]..In this system computer vision and alcohol gas sensor combination is used to detect drowsiness and alcohol intoxication. This system makes use of Raspberry-pi and Arduino UNO with I2C protocol. The proposed system is based on computer vision and embedded system applications. Eye closure is detected using HAAR based cascade classifier and an alcohol gas sensor which functions as a Breathalyzer [4]. This system includes two modules. The two modules are the face and eye detection module followed by the face tracking module. CAMSHIFT algorithm is used for continuous face tracking. This system also uses cascade classifiers in order to improve the accuracy of face detection. The system is a real time non- intrusive model [8]. To reduce the number of accidents caused by drowsiness, various methods for detecting drowsiness automatically have been developed. Three ideas are discussed in this paper; the first idea is creating a dataset of drowsy facial expressions. The second idea is to combine

      3. Adjust centre of the window to mass centre.

      4. Repeat 2 and 3 until distance of the two centres (centre of the window and the mass centre) is less than some threshold value.

    3. HAAR TRAINING

      The OpenCV library provides numerous functions for face and feature (eyes, mouth, sunglasses, etc) detection. Some of these functions can be used to train classifiers. The classifiers can be trained for the process of detection of face This is known as HAAR training. Object. Here, a cascade function is trained from a number of images, both positive and negative. Each feature is a single value obtained by subtracting sum of pixels under various regions of the images [3]. The pixels used for extraction is different for each feature. All the extracted features will not be useful for the required process.

      The Adaboost technique is used to extract the relevant features. Each and every feature is applied on the training images. The best threshold is determined for each feature which classifies images from positive to negative. Features which provide the least error rate are chosen. Initially each feature is given an equal weight. As the process continues, the weights are updated according to the results obtained in

      order to improve the accuracy. The weighted sum of the weak classifiers is the final classifier.

    4. VIOLA JONES ALGORITHM

      Viola Jones algorithm uses the following techniques in its algorithm [8]. They are:

      • HAAR based features

      • Integral Image Formation

      • AdaBoost Technology

      • A cascade of classifiers

    Features are selected based on the pixel intensities in HAAR based feature representation. It does not take into consideration, the values of the pixel. HAAR based features are scalar product between the image and some HAAR templates.

    Integral image formation is used for feature calculation. It considers only four corners of the image. Adaptive boosting (AdaBoost) is used to select the required features. Due to the use of Adaptive Boosting there is a reduction in the computational time of the algorithm.

    A cascade of classifiers is used in order to develop a strong classifier chain. The OpenCV library provides a command prompt training utility called HAAR-training which generates a classifier in XML format when given positive and negative examples of the object to be detected.

  4. CONCLUSION

This paper provides a comparative study on papers related to driver drowsiness detection and alert system.

In order to provide a solution to the problem of detecting the state of drowsiness, an arithmetic based method is used [2]. This system uses eye movement in order to detect fatigue. Eye movement is detected using a camera. This is done to recognize the symptoms of fatigue in order to avoid accidents. [6]. It is based on the concept of eye-tracking. In order to obtain finer results, a hundred and fifty images of different people have been used. If the state of fatigue has been identified, an alarm system is turned on [9]. Computer vision with embedded systems are used. A software algorithm is developed. It was partially tested and found to be effective. There is much scope for further improvements [4]. The proposed system detects drowsiness if the eyes have been closed for a period of four or more frames. The detection system differentiates the normal eye blink from drowsiness. The developed system is a non-invasive system. The system can be further developed by adding various types of sensors [5].The system is based on

computer vision. It makes use of the Viola Jones algorithm, AdaBoost classifier and CAMSHIFT algorithm. A low-cost application can be devised by implementing this system using a raspberry-pi module [8]. The main aim of the paper is to develop a software tool detection of the state of fatigue. It was found to be a timely and accurate technique. Here, input is captured by a camera, processed by the Raspberry-pi module, and the output is in the form of a buzzer that alerts the user, as and when, drowsiness is detected [3]. It groups drowsiness detection techniques into two kinds, driver based and vehicle based. It also provides a survey of numerous driver and vehicle-based techniques [11]. This system is based on the shape predictor algorithm. It provides a non-intrusive approach for drowsiness detection. In future, the frequency of yawning can also be used as a parameter to detect drowsiness [7]. In order to detect drowsiness, certain facial features were identified. This system uses the concept of video processing. It also mentions certain disadvantages of the proposed system and methods to overcome those disadvantages [1].

REFERENCES

  1. Kyong Hee Lee, Whui Kim, Hyun Kyun Choi, Byung Tae Jan. A Study on Feature Extraction Methods Used to Estimate a Drivers Level of Drowsiness, IEEE, February 2019.

  2. Tianyi Hong, Huabiao Qin, Drivers Drowsiness Detection in Embedded System., IEEE, December 2007.

  3. Lorraine Saju, Christeena J, Farhana Yasmin, Surekha Mariam, Drowsiness detection system for drivers using HAART training and template matching, IJEAST, Vol. 1, Issue 6, April 2016.

  4. Dwipjoy Sarkar, Atanu C, Real Time Embedded System Application for Driver Drowsiness and Alcoholic Intoxication Detection, IJETT, Volume 10 Number 9, April 2014.

  5. SrinivasuBatchu, S Praveen Kumar, Driver Drowsiness Detection to Reduce the Major Road Accidents in Automotive Vehicles, IRJET, Volume 02 Issue 01, April 2015.

  6. Hardeep Singh, J S Bhatia and Jasbir Kaur, Eye Tracking based Driver Fatigue Monitoring and Warning System, IEEE, January 2011.

  7. Fouzia, Roopalakshmi R, Jayantkumar A Rathod, Ashwitha S, Supriya K, Driver Drowsiness Detection System Based on Visual Features. , IEEE, April 2018.

  8. Varsha E Dahiphale, Satyanarayana R, A Real-Time Computer Vision System for Continuous Face Detection and Tracking, IJCA, Volume 12 Number 18, July 2015.

  9. SaeidFazli, Parisa Esfehani, Tracking Eye State for Fatigue Detection, ICACEE, November 2012. Gao Zhenhai, Le DinhDat, Hu Hongyu, Yu Ziwen, Wu Xinyu, Driver Drowsiness Detection Based on Time Series Analysis of Steering Wheel Angular Velocity, IEEE, January 2017.

  10. Bagus G. Pratama, IgiArdiyanto, Teguh B. Adji, A Review on Driver Drowsiness Based on Image, Bio-Signal, and Driver Behavior, IEEE, July 2017.

Leave a Reply

Your email address will not be published. Required fields are marked *