Rancang Bangun Sistem Pengendali Lampu Lalu Lintas Berdasarkan Pengenalan Citra Digital Kendaraan Menggunakan Metode Faster R-Cnn

Ma'ali, Ahmad Miqdad and Hendriyawan A, M.S (2019) Rancang Bangun Sistem Pengendali Lampu Lalu Lintas Berdasarkan Pengenalan Citra Digital Kendaraan Menggunakan Metode Faster R-Cnn. Tugas Akhir thesis, University of Technology Yogyakarta.

[img]
Preview
Text
Naskah Publikasi_Ahmad Miqdad M.pdf

Download (750kB) | Preview

Abstract

ABSTRACT Traffic jam is a common thing that happen nowadays, especially in big cities. Traffic jam is mostly caused by the increasing of vehicles amount in the roads. It causes traffic jam, especially in a cross road, where the intensiveness of every lane is not the same. For example, in one lane there are lots of vehicles that caused traffic jam and the other lane is deserted, so that the crowd will be more crowded and the deserted will be more deserted. This happens because the duration of green lamps are the same. To manage the fluency of vehicles on the roads it requires the adaptive traffic lamp based on level of intensiveness on every lane. This day, adaptive traffic lamp management is based on time prediction of traffic jam. It is not so effective because the traffic jam time is not always same every day. Solution provided in this research to implement adaptive traffic jam is using computer vision and digital image processing so that traffic lamp is adjusted for real time through the visual. Using faster r-cnn method to detect vehicle objects in real-time of every lane, this system can detect through the trained vehicle dataset. System will count the amount of vehicles on the road to determine duration of green and red lamp based on the level of intensiveness. The result of this research is to detect and count vehicles in a cross road simulation using custom dataset of vehicle toys to determine the traffic lamp duration. Accuration percentage of this dataset is 97.027% and the percentage of error calculation of amount vehicles is 2.188%. Key word : Computer Vision, Digital Image Processing, Dataset, Deep Learning, Faster R-Cnn

Item Type: Thesis (Skripsi, Tugas Akhir or Kerja Praktek) (Tugas Akhir)
Subjects: T Technology > TK Electrical engineering. Electronics Nuclear engineering
Divisions: Fakultas Teknologi Informasi dan Elektro > S1 Teknik Elektro
Depositing User: UTY MS Hendriyawan Achmad
Date Deposited: 02 Aug 2019 04:22
Last Modified: 02 Aug 2019 04:22
URI: http://eprints.uty.ac.id/id/eprint/3335

Actions (login required)

View Item View Item