<![CDATA[Traffic accident is one of the most significant contributors which makes the death number is increasing around the world. With the demographic condition from Indonesia, motorcycle driver is the types of the driver that dominated the traffic, therefore increasing the probability of caught in a traffic accident The existing Vehicle Activity Detection System (VADS) mainly focused on the car driver, with the main problem is that the computational time from the system is too high to be implemented on a real-time condition. To solve this problem, in this research, a classification system for abnormal driving behavior from motorcycle drivers is created, using Light Gradient Boosting Machine (LightGBM) model. The system is designed to be lightweight in computation and very fast in response to the changes of the activities with a high velocity. To train the LightGBM model, the data from Accelerometer and Gyroscope sensor, that has been integrated into a smartphone, will be used to detect the movement from a driver. The accuracy rate from the proposed model is reaching 82% on the test dataset and shows a promising result of around 70% on the real-time detection process. With a computational time of around 10ms, the proposed system is able to work 5 times faster than the existing system.]]>
Copyrights © 2021
