ICoSI 2019 Conference

The Modification of Vehicle Detection Program Based on Java by Using Gaussian Mixture Model (GMM) Method for Rural Street
Muchlisin, Nurtia Amanda

Department of Civil Engineering, Faculty of Engineering, UMY, Indonesia


Population growth in Indonesia is currently experiencing a high increase. Those things will affect the increase of transportation needs, so that might increase the traffic volume. Traffic volume is needed in transportation planning. Nowadays, the manual method is usually used by field observers using a calculation tool called a counter to get traffic volumes. With the high of traffic volumes, manual traffic volume calculations are less effective. For this reason, in this study the researchers modified the program to create a vehicle detection program and classify vehicles according to MKJI 1997 on a Java-based off-street called U-COUNTER (Rural Area). The method that used in this research is Gaussian Mixture Model (GMM) method that base in image processing. In this program will attack of video input in the morning with 6 m height and 7 m, as well as video input in the afternoon with 6m and 7m height. The calculation results on the program will compared with manual calculations to get the accuracy and percent error values. The highest accuracy results obtained in this program are 76.48% with an error value of 23.52% for video 6 m in the morning. For the lowest accuracy results found on the 7 m video in the afternoon with an accuracy of 43% with an error value of 57%, this caused by congested traffic during video capture.

Keywords: Gaussian Mixture Model (GMM), Rural Street, Image Processing, Traffic Volume.

Topic: International Symposium of Civil, Environmental, and Infrastructure Engineering

Link: https://ifory.id/abstract-plain/EavpxWrgcBjV

Web Format | Corresponding Author (Muchlisin ST, M.Sc.)