Sistem Deteksi Kantuk Pada Pengendara Roda Empat Menggunakan Eye Blink Detection

Abstract

The number of traffic accidents in Indonesia is increasing. One of the main factors is the condition of a sleepy driver. In general, sleepy occurs at night when the body need to take rest. But in some people sleepy does not appear to depend on time. This situation needs more attention if we are driving so that the accident  cause of sleepy can be avoided. The Republic of Indonesia Police noted that the traffic accidents increasing continued throughout the year, where almost the majority occurred because of drivers who were in a sleepy condition. From these problems a system is created that can automatically determine whether the driver is in  conscious, sleepy or a sleep using the Haar Cascade Classifier method. The process begin with taking pictures using Pi Camera which is connected to raspberry to detect the face area using the Haar Cascade Classifier method then the regression tress algorithm on facial landmarks detection which is used to detect sleepy eyes with the output of an alarm to react so the driver is not sleepy. From the results of the overall trial conducted, the percentage of success was 90% and the error rate was 10% during the day from 20 experiments. While the percentage of testing at night obtained a value of 85% and an error rate of 15% from 20 experiments. With the distance between the camera and the driver between 30-50 cm and the slope of 0 - 45 degrees