Abstract: The human body requires nearly 7 to 8 hours of sleep. But due to work stress it is lacking. Due to a lack of sleep and irregular sleep cycles, humans tend to feel drowsy at any time of the day. Hence with these poor work-life timings, people may find it difficult to perform tasks such as driving that require mental and physical health and well-being. Sleepless driving can cause problems and therefore it is necessary to be awake while driving. Due to the drowsiness of drivers, accidents are caused which kill thousands of people worldwide every year. Therefore there is a need for a sleep sensor application to help prevent such accidents and save lives. In this regard, we propose a system based on neural networks (CNN) and a Haar cascade classifier with a user interface to deal with the problem. In this research we aim to develop a prototype "Real-time drowsiness detection system". The system monitors the driver's eyes and rings the alarm while he is drowsy. This system is object oriented, which is built using OpenCV library in python where the images are gathered from the webcam and fed into a Haar based algorithm which will detect face and eyes and later a CNN model which will predict whether the person’s eyes are ‘Open’ or ‘Closed’. A web application for this model with a user interface and also a CSV file to store the details of the user is developed. Experiments were conducted to test the effectiveness of the proposed method in comparison to other methods. The empirical results show that the proposed method using deep learning techniques can achieve a high accuracy of 98.5%. This study provides a solution to prevent automobile accidents due to drowsiness.

Keywords: Deep Learning, Convolutional Neural Network (CNN), Haarcascade, Keras, Drowsiness detection, Eye’s detection

| DOI: 10.17148/IARJSET.2022.9122