Steven, Steven (2020) Penambahan image-library YOLO untuk meningkatkan akurasi dalam identifikasi kendaraan di jalan tol Indonesia = Addition of image-library YOLO to increase accuracy in vehicle identification in Indonesia toll road. Bachelor thesis, Universitas Pelita Harapan.
![Title [thumbnail of Title]](http://repository.uph.edu/style/images/fileicons/text.png)
Title (3).pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (1MB)
Preview
ToC.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (1MB) | Preview
Preview
Abstract.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (191kB) | Preview
Preview
Chapter1.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (373kB) | Preview
![Chapter2 [thumbnail of Chapter2]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter2.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (2MB)
![Chapter3 [thumbnail of Chapter3]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter3.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (2MB)
![Chapter4 [thumbnail of Chapter4]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter4.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (4MB)
![Chapter5 [thumbnail of Chapter5]](http://repository.uph.edu/style/images/fileicons/text.png)
Chapter5.pdf
Restricted to Registered users only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (237kB)
Preview
Bibliography.pdf
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (306kB) | Preview
![Appendices.pdf [thumbnail of Appendices.pdf]](http://repository.uph.edu/style/images/fileicons/text.png)
Appendices.pdf
Restricted to Repository staff only
Available under License Creative Commons Attribution Non-commercial Share Alike.
Download (374kB)
Abstract
Jalan tol dibuat agar orang dapat terhindar dari kemacetan, tetapi beberapa kali seorang pekerja yang pulang pergi setiap hari dapat menghadapi beberapa kejadian seperti: kecelakaan, kendaraan mogok, dan kepadatan lalu lintas. Untuk mengetahui kondisi jalan tol, P.T. Jasa Marga telah memasang CCTV. CCTV ini akan menampilkan kondisi jalan tol secara langsung. Jadi, petugas dapat berbagi informasi tentang lalu lintas melalui media elektronik seperti televisi, radio atau telepon. Hampir semua jalan tol di Indonesia memiliki CCTV yang dipasang di sisi atau di tengah jalan. P.T. Jasa Marga sebagian besar telah menyediakan foto beresolusi rendah dan video 3 detik, tetapi tidak dapat mendeteksi kemacetan lalu lintas, memberikan informasi kecepatan atau menghitung jumlah kendaraan di jalan tol tertentu, maupun memberikan data kepadatan jalan tol. Dalam kasus-kasus tertentu, misalnya selama situasi darurat seperti situasi pandemi saat ini, data ini bahkan lebih penting, karena penumpang harus mengubah kebiasaan transportasi publik mereka menjadi kendaraan pribadi.
YOLO adalah algoritma untuk mendeteksi objek. YOLO dapat mendeteksi berbagai jenis objek, mulai dari orang, mobil, truk, pesawat, dan lain-lain. YOLO harus dilatih untuk membuat kumpulan data untuk tujuan deteksi. Hasil pelatihan dapat meningkatkan deteksi di jalan tol P.T. Jasa Marga. Dalam laporan ini, YOLO didukung oleh Darkflow untuk mendeteksi objek. Ada tiga proses yang akan dilakukan, yaitu proses pembuatan label, proses pelatihan, dan proses deteksi. Proses pelabelan menggunakan labelImg, sementara pelatihan dan proses deteksi menggunakan Darkflow. Ketika ketiga proses ini telah dilakukan secara lengkap, proses ini dapat diterapkan di jalan tol Indonesia. Dalam laporan ini, kami telah menyediakan YOLO-image-library yang lebih baik sehingga deteksi bisa lebih akurat daripada menggunakan perpustakaan gambar YOLO asli.
Dari penelitian yang dilakukan, hasil pendeteksian data YOLO dari proses pelatihan menggunakan gambar jalan tol Indonesia yang diambil oleh CCTV P.T. Jasa Marga lebih akurat sebesar 44% dibandingkan dengan hasil deteksi data YOLO asli.
/
Toll roads are made so that people can avoid traffic delays, but some times a commuter can face several incidents such as: accidents, stalled vehicles, and congestions, which will cause traffic jam. To find out the condition of many toll roads, P.T. Jasa Marga has installed CCTV. These CCTV will display the conditions of toll roads directly. So, officers can share information about the traffic through electronic media such as television, radio or telephone. Almost all toll roads in Indonesia have CCTV installed on their sides or in the middle. P.T. Jasa Marga has provided mostly low resolution photos and 3 second videos, but could not detect traffic jams, provide speed information nor count the number of vehicles on a certain toll road, let alone provide density data. In certain cases, for example during emergency situation like present pandemic situation these data are even more important, since commuters have to change their public transport habit in to individual/private vehicles.
YOLO is an algorithm for detecting objects. YOLO can detect various types of objects, ranging from people, cars, trucks, planes, etc. YOLO must be trained to create a data set for the purpose of detection. The results of the training can improve detection on the toll road of P.T. Jasa Marga. In this report, YOLO is supported by Darkflow to detect objects. There are three processes that will be carried out, namely the process of creating labels, training process, and detection process. The labeling process uses labelImg, while training and the detection process uses Darkflow. When all three of these processes have been carried out more completely, they can be applied on the Indonesian toll roads. In this report, we have provided a better YOLO-image-library so that detections can be more accurate than using the original YOLO image library.
From the research conducted, the results of YOLO data detection from the training process using Indonesian toll road images taken by CCTV P.T. Jasa Marga are more accurate by 44% compared to the results of the original YOLO data detection.
Item Type: | Thesis (Bachelor) |
---|---|
Creators: | Creators NIM Email ORCID Steven, Steven NIM00000011580 s00000011580@student.uph.edu UNSPECIFIED |
Contributors: | Contribution Contributors NIDN/NIDK Email Thesis advisor Hardjono, Benny NIDN0404086401 benny.hardjono@uph.edu Thesis advisor Prasetya, Kusno NIDN0718087903 kusno.prasetya@uph.edu |
Additional Information: | SK 82-15 STE p |
Uncontrolled Keywords: | YOLO; Darkflow; deteksi objek; Python; sedan; truk |
Subjects: | Q Science > QA Mathematics > QA75 Electronic computers. Computer science |
Divisions: | University Subject > Current > Faculty/School - UPH Karawaci > School of Information Science and Technology > Informatics Current > Faculty/School - UPH Karawaci > School of Information Science and Technology > Informatics |
Depositing User: | Steven Steven |
Date Deposited: | 10 Aug 2020 01:50 |
Last Modified: | 04 Aug 2021 08:01 |
URI: | http://repository.uph.edu/id/eprint/9610 |