| dc.description.abstract | Driver fatigue and drowsiness contribute to more than 20% of reported road accidents
worldwide, with motorcyclists being particularly vulnerable. To address this issue, this
study proposes a modified helmet with innovative features aimed at detecting and prevent ing rider drowsiness. To estimate driver drowsiness, the study utilizes various techniques,
such as monitoring biomedical signals, visually assessing the driver’s bio-behaviour through
facial images and observing the driver’s performance. The proposed algorithm focuses
on live monitoring of the Eye Aspect Ratio (EAR) using image processing techniques.
High-definition live video is decomposed into continuous frames, and facial landmarks
are detected with a pre–trained neural network based on Dlib functions, trained using the
HAAR Cascade algorithm. The image processing library, OpenCV, plays a key role in this
algorithm’s implementation, which is carried out in Python. By calculating the EAR and
continuously monitoring it against a predetermined threshold value, the algorithm can
detect blinks and micro-sleep episodes. The detected blinks and the level of drowsiness
are displayed on the monitor screen, accompanied by a vibration warning for micro sleep detection. In conclusion, this study presents an effective algorithm that leverages
live monitoring of the EAR through image processing techniques to estimate driver
drowsiness. Its implementation demonstrates promising results in identifying blinks,
assessing drowsiness levels, and providing timely warnings to mitigate the risks associated
with driver fatigue and drowsiness, thereby enhancing road safety for motorcyclists | en_US |
