Limitations of CNN-Model and Enhanced AI- Model for Driver Drowsiness Detection

Abstract

Abstract: Contemporarily driver drowsiness is one of the major universal facts of road accidents across the globe. Integrating enhanced Information Technology (IT) for instance Neural Network (NN), Computer Vision (CV), and Image Processing (IP), enormously reduce road accidents from drivers’ drowsiness while driving. Advanced computer vision technology, artificial neural network, and intelligent cameras dynamically predicate as well as alert the drivers when they are in drowsiness. Drowsiness Detection (DD) has been an important research domain in real-time biomedical and traffic signal applications. Recently various Deep Learning (DL) algorithms implemented to study and diagnose fatigue situations in Electroencephalograms Signals (EEGs). The primary objective proposed in the present article is to study a survey of literature on drivers’ drowsiness detection based on driver behavioral measures by using computer vision and artificial neural network algorithms.   Furthermore, in this article, traditional drivers’ drowsiness architecture models, limitations, challenges, and conventional neural network algorithms have been addressed. As regards to study also addressed the role of enhanced convolution neural networks and limitations in the present and future scenario to detect and alert drivers’ drowsiness during driving which leads to reduced road accidents in the future globally.

Keywords: Computer vision, convolutional neural network, drowsiness detection, deep learning, and information technology