Infrastructure has seen tremendous growth globally. There is an increase in the number of vehicles commuting from one place to another. The increase in vehicles has also seen a gradual increase in the number of road crashes annually. The World Health Organization (WHO) has declared road traffic injuries and deaths as a severe global problem. 2015 accounted for 1.25 million deaths due to car crashes, and 2018 accounted for 1.35 million deaths. 23.5% of all the accidents are due to drivers driving in a state of drowsiness. In this paper we propose two models. The first model can classify whether a driver is drowsy or not. The model achieved an accuracy of 92.40%. The second model is a multi-class model which can identify the state of the driver such as yawning, blinking, talking etc. A total of 12 states can be identified by the model. Multitask learning encourages features in one sub task to be used in classification for other sub tasks. Using multitask learning, indicators of drowsiness in a driver is identified.
Detection and State Analysis of Drowsiness using Multitask Learning with Neural Networks
21.10.2020
445011 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Driver Drowsiness Detection Using Deep Learning
IEEE | 2023
|Driver Drowsiness Detection Using Deep Learning
IEEE | 2021
|Driver Drowsiness Detection using Deep Learning
IEEE | 2022
|Driver Drowsiness Detection Using Machine Learning
IEEE | 2023
|