DETEKSI JENIS KENDARAAN BERMOTOR DENGAN ALGORITMA DETECTION TRANSFORMER (DETR)

APRIAN, FERDI and Sutarno, Sutarno (2025) DETEKSI JENIS KENDARAAN BERMOTOR DENGAN ALGORITMA DETECTION TRANSFORMER (DETR). Undergraduate thesis, Sriwijaya University.

	Image RAMA_56201_09011282126099_cover.jpg - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (424kB)
	Text RAMA_56201_09011282126099.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (5MB) | Request a copy
	Text RAMA_56201_09011282126099_TURNITIN.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (8MB) | Request a copy
	Text RAMA_56201_09011282126099_0201117802_01_front_ref.pdf - Accepted Version Available under License Creative Commons Public Domain Dedication. Download (1MB)
	Text RAMA_56201_09011282126099_0201117802_02.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (925kB) | Request a copy
	Text RAMA_56201_09011282126099_0201117802_03.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (549kB) | Request a copy
	Text RAMA_56201_09011282126099_0201117802_04.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (1MB) | Request a copy
	Text RAMA_56201_09011282126099_0201117802_05.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (268kB) | Request a copy
	Text RAMA_56201_09011282126099_0201117802_06_ref.pdf - Bibliography Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (242kB) | Request a copy
	Text RAMA_56201_09011282126099_0201117802_07_lamp.pdf - Accepted Version Restricted to Repository staff only Available under License Creative Commons Public Domain Dedication. Download (825kB) | Request a copy

Abstract

Increased traffic in the city of Palembang has caused problems such as congestion and accidents, necessitating an accurate vehicle detection system. This study proposes the use of DETR and RT-DETR to identify vehicle types in traffic images, with ResNet-50 and ResNet-101 as comparison architectures. The dataset consists of 1,233 images extracted from traffic videos in Palembang. The model was trained using PyTorch Lightning and a GPU for computational efficiency. Evaluation was conducted using AP, mAP, AR, and mAR metrics. The results show that RT-DETR with the ResNet-101 backbone and a batch size of 4 provides the best performance, with mAP of 0.558 and mAR of 0.221. This study demonstrates that architecture selection significantly impacts accuracy and can serve as a foundation for the development of intelligent transportation systems in the future.

Item Type:	Thesis (Undergraduate)
Uncontrolled Keywords:	Vehicle Detection, Detection Transformer (DETR), Real-Time Detection Transformer (RT-DETR), ResNet.
Subjects:	Q Science > Q Science (General) > Q334-342 Computer science. Artificial intelligence. Algorithms. Robotics. Automation.
Divisions:	09-Faculty of Computer Science > 56201-Computer Systems (S1)
Depositing User:	Ferdi Aprian
Date Deposited:	15 Aug 2025 01:41
Last Modified:	15 Aug 2025 01:41
URI:	http://repository.unsri.ac.id/id/eprint/182773

Actions (login required)

View Item