<![CDATA[Traffic density can be controlled by obtaining and managing the data of the traffic flows on the highway. Generally, the process of data acquisition of the traffic flows which passing on the highway are still done manually by assigning some officers to be on the highway and count each of passing vehicle, then divided by a certain time frame. This manual counting are still have many weaknesses such as time of collecting data become longer, and need much amount of the human resources. Based on these conditions, needs an accurate automatic vehicle detection and counting system as traffic monitors, traffic controllers and traffic analysis. At this time, it has been developed a vehicle detection system using a hardware system such as using sensors, Radio Frequency Identifier or other hardware which integrated by software in the microcontroller and works automatically to detect the speed and count the number of passing vehicles on the highway. The weaknesses of these detectors can only detect at the narrow range, design of the system, the complexity of the operation, and also has a significant operational cost. Based on those system weaknesses, this study was developed with a focus of designing the detection system and the vehicle counter system using Kalman filter and Gaussian Mixture Models (GMM) method. This system get the most accurate results in the morning (10,000-25,000 lux illumination) with F1 Score value of 0.91111, while counting the vehicles most inaccurate happen at night (illumination from 0.27 to 1.0 lux) with F1 Score only 0.16071.Keywords: Traffic flows data, vehicle counter system, Object Tracking, Gaussian Mixture Model, Kalman Filter.]]>
