Driver Drowsiness Detection Using Visual Information On Android Device

摘要

The number of casualties from road accidents keep arising each year. While there are many causes to road accidents, most are surprisingly caused by human errors, such as drowsiness. Therefore, this issue raises an idea for an application which will be able to track a person's eye movement while driving. The application, called “Driver Drowsiness Detection”, runs on an Android handheld and wearable. The purpose of this application is to alert drivers so that they can be cautioned to pull over and stop driving in a drowsy state. The application “Driver Drow-siness Detection” utilizes Haar-cascade Detection as well as template matching in OpenCV to detect and track the eyes using the front camera of an Android device. Testing has been conducted to ensure that the functionality, behavior, performance and user satisfaction are as expected. Even though, the input received by the application still has several restrictions, specifically in correlation to sufficient lighting and obscurity of the face and eyes area, the application has successfully detected the eye blinks at the angle of 30 to 60 degrees and distance of 20-50 cm, as well as measuring the heart rate.