Drowsiness and microsleep while driving are highly dangerous as they reduce alertness and trigger fatal accidents. This study aims to develop an early drowsiness detection system to improve safety and reduce the risk of traffic accidents. The method utilizes MediaPipe Face Mesh and Eye Aspect Ratio (EAR) operating in real-time. The system extracts video from a camera to detect facial landmarks and calculate the eye closure ratio. If the EAR value falls below the 0.18 threshold for more than 1 second, the system automatically triggers a warning alarm. Evaluations were conducted through 24 direct tests (12 daytime and 12 nighttime conditions) as well as testing on a driving simulation video. The test results demonstrated an accuracy rate of 100% during the day, 91.66% at night, and 81.82% for the simulation video. In conclusion, this system is proven to be effective and reliable in detecting driver drowsiness under various lighting conditions, thus holding great potential for implementation to prevent occupational and traffic accidents.
Copyrights © 2025