PENDETEKSIAN STIKER DAN KARAKTER PADA PELAT NOMOR UNTUK SISTEM PARKIR TANPA KARCIS DENGAN MEMANFAATKAN OPTICAL CHARACTER RECOGNITION DAN MODEL YOU ONLY LOOK ONCE (YOLO)

Abstract

Parking systems in general use one of two systems: using a ticket or not using a ticket when a vehicle wants to pass through the parking portal. A ticketless parking system is not practical because it requires parking attendants to give a ticket to visitors and collect data by recording plate numbers manually or by hand. Nonetheless, the distinction between general and special visitors is needed in the parking system for some specific cases. This research proposes the utilization of stickers to enhance information about special visitors and anticipate when character recognition errors happen on vehicle license plates. Optical Character Recognition (OCR) and the You Only Look Once (YOLO) model are used in this research to detect stickers and characters on vehicle license plates. Based on the results of testing the impact of distance on the performance of detecting character and sticker together, the character recognition accuracy rate reached 82.3% at a distance of 30 cm, 91.1% at a distance of 60 cm, and 86.1% at a distance of 100 cm, while the sticker recognition performance was 100% for all distances.

Keywords: OCR, YOLO, object detection, parking system, distance.