Car crashes are a major reason people get hurt or die. Every year, about a million people worldwide lose their lives in these accidents, according to the World Health Organization. One of the biggest dangers is drivers being too tired. When people don't get enough sleep or are just really worn out, they can fall asleep while driving, putting themselves and others at risk. Studies have shown that being sleepy is a main cause of serious car accidents. These days, it seems like driving when you're tired is the biggest reason drivers get sleepy. And this sleepiness is causing more and more accidents. This is a big problem that we need to solve quickly. Many people are trying to create devices that can tell when a driver is getting sleepy, right when it is happening. These devices use different kinds of artificial intelligence. Our research is also about finding ways to tell when a driver is sleepy. We are building a system that looks at the driver's face and tracks their eyes. It takes pictures of the driver's eyes and compares them to a set of example pictures. If the system sees that the driver's eyes are closed for too long, it sounds an alarm to wake them up. If the driver opens their eyes, the system keeps watching. We have a score that goes up when the eyes are closed and down when they are open. In this paper, we will explain how we built a system that can detect drowsiness with 80% accuracy. Our goal is to help make roads safer by reducing accidents caused by tired drivers.

Face Detection, Euclidean Eye Aspect Ratio, Electrooculography, CNN, Alarm, Eye blinking

In summary, driver drowsiness detection systems represent a crucial vehicle safety technology designed to mitigate injuries caused by fatigued driving. The timely detection and alerting of drivers experiencing drowsiness are paramount in preventing potentially fatal accidents. The proposed system utilizes image processing techniques to assess driver drowsiness levels by quantifying the Eye Aspect Ratio (EAR), effectively measuring the driver's eye aperture. To establish a reliable drowsiness threshold, comprehensive EAR data collection is essential. An auditory alarm system serves as a vital component, aiming to reduce the incidence and severity of accidents attributed to driver fatigue, thereby contributing to a decrease in annual vehicle crash fatalities. The current system demonstrates consistent drowsiness detection for individual drivers with minimal limitations. The alarm mechanism functions effectively, providing timely alerts. However, the threshold for triggering the alarm, based on EAR, may vary significantly across individuals. Future research should focus on developing an adaptive threshold determination mechanism that eliminates the need for individual calibration. This would enable the system to automatically establish personalized drowsiness thresholds based on real-time driver behaviour. Furthermore, some drivers may prefer a more sensitive and frequent alarm system due to heightened awareness of road safety. Integrating user-adjustable sensitivity settings could address this variability in driver preferences.

-