Aplikasi Computer Vision Object Detection Pada Pemantauan Ruang Parkir Otomatis Berbasis YOLOv8

Seprianto, Pras (2025) Aplikasi Computer Vision Object Detection Pada Pemantauan Ruang Parkir Otomatis Berbasis YOLOv8. Sarjana (S1) thesis, Universitas Islam 45 Bekasi.

	Text PENDAHULUAN.pdf Download (1MB)
	Text BAB I.pdf Restricted to Repository staff only Download (150kB) | Request a copy
	Text BAB II.pdf Restricted to Repository staff only Download (243kB) | Request a copy
	Text BAB III.pdf Restricted to Repository staff only Download (181kB) | Request a copy
	Text BAB IV.pdf Restricted to Repository staff only Download (353kB) | Request a copy
	Text BAB V.pdf Restricted to Repository staff only Download (141kB) | Request a copy
	Text DAFTAR PUSTAKA.pdf Download (75kB)

Abstract

The rapid growth of motor vehicles has led to an imbalance between the number of vehicles and the availability of parking spaces, particularly in urban areas. One potential solution to address this issue is the implementation of an automated parking space monitoring system based on computer vision. This study aims to compare the performance of the YOLOv8 object detection algorithm using two different model variants (library N and library M) in detecting car objects, as well as to evaluate its ability to distinguish cars from other types of objects. The research employs a quantitative method, where accuracy and performance are measured using the formulas for TP, FP, FN, precision, recall, and mAP@50. These metrics serve as provisional answers to the research questions by running a computer vision program to detect objects in CCTV video footage. The recorded videos are extracted into individual frames for manual testing, allowing a detailed analysis of which objects are successfully detected by YOLOv8 and determining which model N or M performs better. The results show that library M achieves higher values in precision, recall, mAP@0.5, and mAP@0.5:0.95 compared to library N, indicating superior object detection accuracy. Although library N demonstrates an advantage in processing speed (faster by 5.6 milliseconds per frame), library M still maintains an efficient processing time, averaging 10.2 milliseconds per frame. Further testing using videos containing non-car objects also showed that library M maintains high performance, with a precision of 1.0 and a recall of 0.993 for the car class. This demonstrates that the YOLOv8 algorithm using library M can accurately and consistently distinguish car objects from others, making it an effective solution for automated parking space monitoring systems. Keywords: YOLOv8, Object Detection, Computer Vision, Parking Monitoring.

Actions (login required)

View Item