Sanni Deslia Pasaribu, - (2023) IMPLEMENTASI ALGORITMA YOLOv8 DAN DEEP LEARNING UNTUK RANCANG BANGUN SISTEM DETEKSI OBJEK KENDARAAN PADA JAM SIBUK DI JALAN PASIR KALIKI (PASKAL TIMUR)-TJOKROAMINOTO BANDUNG. S1 thesis, Universitas Pendidikan Indonesia.
![[img]](http://repository.upi.edu/style/images/fileicons/text.png) |
Text
S_SISTEL_1903119_Title.pdf Download (271kB) |
|
Text
S_SISTEL_1903119_Chapter1.pdf Download (249kB) |
|
Text
S_SISTEL_1903119_Chapter2.pdf Restricted to Staf Perpustakaan Download (170kB) |
|
Text
S_SISTEL_1903119_Chapter3.pdf Download (420kB) |
|
Text
S_SISTEL_1903119_Chapter4.pdf Restricted to Staf Perpustakaan Download (576kB) |
|
Text
S_SISTEL_1903119_Chapter5.pdf Download (74kB) |
|
Text
S_SISTEL_1903119_Appendix.pdf Restricted to Staf Perpustakaan Download (379kB) |
Abstract
Deteksi objek pada kendaraan menggunakan implementasi algoritma YOLO sudah sering ditemui pada berbagai penelitian. Tujuan dari penelitian semacam ini juga beragam-ragam pengimplementasiannya, ada untuk deteksi objek berupa buah-buahan, manusia, kendaraan, hewan, dan masih banyak lagi. Dalam penelitian ini juga membahas terkait implementasi YOLOv8 untuk deteksi objek dengan arsitektur CNN untuk mendeteksi jenis kendaraan yang dibagi menjadi 3 class, yaitu mobil, motor, dan kendaraan besar. Lalu dibangun sistem traffic counting untuk menghitung jumlah kendaraan melalui video rekaman CCTV yang ada di Jalan Pasar Kaliki (Paskal Timur)-Tjokroaminoto, Bandung. Berdasarkan hasil sistem yang dirancang, dapat dikatakan sistem berhasil untuk mendeteksi objek dan menghitung jumlah kendaraan di waktu yang ditentukan. Hasil training data yang diperoleh mencapai hasil presisi (mAP) mencapai 93,5%, box loss 0,15 dan class loss 0,104 memiliki nilai yang mendukung untuk dibangunnya sistem. Hasil demo sistem yang sudah berhasil dirancang dengan algoritma YOLOv8 dan Deep Learning, mendapatkan data jumlah kendaraan selama 1 minggu dengan jumlah kendaraan motor lebih banyak dengan total 55.473 ribu, mobil, 27.407 ribu dan kendaraan besar sebanyak 849. ----- Object detection in vehicles using the implementation of the YOLO algorithm is frequently encountered in various research studies. The objectives of such research are diverse in their implementations, ranging from object detection for fruits, humans, vehicles, animals, and much more. This study also discusses the implementation of YOLOv8 for object detection using a CNN architecture to identify types of vehicles, which are categorized into 3 classes: cars, motorcycles, and large vehicles. Furthermore, a traffic counting system is developed to tally the number of vehicles through CCTV video recordings on the road from Pasar Kaliki (East Paskal) to Tjokroaminoto Street in Bandung. Based on the outcomes of the designed system, it can be concluded that the system successfully detects objects and counts the number of vehicles within the specified timeframe. The training results achieved a precision (mAP) of 93.5%, with a box loss of 0.15 and a class loss of 0.104, indicating supportive values for the system's construction. The demonstration of the system, designed using YOLO and Deep Learning algorithms, yielded data on vehicle counts for a week, with motorcycles comprising the majority at a total of 55,473, followed by cars at 27,407, and large vehicles totaling 849.
| Item Type: | Thesis (S1) |
|---|---|
| Additional Information: | Link Google Scholar (Dewi Indriati Hadi Putri): https://sinta.kemdikbud.go.id/subjects/detail/58915 Link Google Scholar (Galura Muhammad Suranegara): https://sinta.kemdikbud.go.id/authors/profile/6703764 ID Sinta Dosen Pembimbing: Dewi Indriati Hadi Putri : 6720737 Galura Muhammad Suranegara: 6703764 |
| Uncontrolled Keywords: | Deteksi objek, YOLO, CNN, traffic counting, prediksi, deep learning. |
| Subjects: | L Education > LC Special aspects of education T Technology > T Technology (General) |
| Divisions: | UPI Kampus Purwakarta > S1 Sistem Telekomunikasi |
| Depositing User: | Sanni Deslia Pasaribu |
| Date Deposited: | 01 Sep 2023 07:02 |
| Last Modified: | 01 Sep 2023 07:02 |
| URI: | http://repository.upi.edu/id/eprint/100588 |
Actions (login required)
 |
View Item |