Driver Safety Development: Real-Time Driver Drowsiness Detection System Based on Convolutional Neural Network
- PDF / 1,418,156 Bytes
- 10 Pages / 595.276 x 790.866 pts Page_size
- 74 Downloads / 218 Views
ORIGINAL RESEARCH
Driver Safety Development: Real‑Time Driver Drowsiness Detection System Based on Convolutional Neural Network Maryam Hashemi1 · Alireza Mirrashid1 · Aliasghar Beheshti Shirazi1 Received: 1 May 2020 / Accepted: 23 August 2020 © Springer Nature Singapore Pte Ltd 2020
Abstract This paper focuses on the challenge of driver safety on the road and presents a novel system for driver drowsiness detection. In this system, to detect the falling sleep state of the driver as the sign of drowsiness, Convolutional Neural Networks (CNN) are used with regarding the two goals of real-time application, including high accuracy and fastness. Three networks introduced as a potential network for eye status classification in which one of them is a Fully Designed Neural Network (FD-NN) and others use Transfer Learning in VGG16 and VGG19 with extra designed layers (TL-VGG). Lack of an available and accurate eye dataset strongly feels in the area of eye closure detection. Therefore, a new comprehensive dataset proposed. The experimental results show the high accuracy and low computational complexity of the eye closure estimation and the ability of the proposed framework on drowsiness detection. Keywords Drowsiness · Driver · Convolutional neural networks · Transfer learning · Safety
Introduction According to published reports from the World Health Organization (WHO), traffic accidents are one of the top 10 causes that lead to death in the world [1]. The reports demonstrate that the first cause of such crashes are drivers. Therefore, the detection of driver drowsiness could be a suitable methodology to prevent accidents. It also improves the performance of the Advanced Driver Assistance Systems (ADAS) and Driver Monitoring System (DMS); as a result, road safety. There are three main categories of drowsiness detectors: Vehicle-based [2], Signal-based [3], and Facial featurebased [4]. Vehicle-based methods try to infer drowsiness from vehicle situation and monitor the variations of steering wheel angle, acceleration, lateral position, etc. However, * Maryam Hashemi [email protected]; [email protected] Alireza Mirrashid [email protected] Aliasghar Beheshti Shirazi [email protected] 1
Iran University of Science and Technology, Resalat highway, Tehran, Iran
these approaches are too slow for real-time tasks. Signalbased methods infer drowsiness from psychophysiological parameters. Several studies have been done during the last years based on these methods [5]. The most critical physiological signals that used and investigated are ElectroEncephaloGram (EEG) [6], ElectroOculoGram (EOG) [7], activities of the autonomous nervous system from ElectroCardioGram (ECG) [8], Skin Temperature (ST), Galvanic Skin Response (GSR) and also intramuscular activities as ElectroMyoGram (EMG). These approaches need to consider invasive captors, which can affect driving negatively. Facial feature-based methods can evaluate the target in realtime without invasive instruments, and they are inexpensive
Data Loading...
