Pengenalan Plat Nomor Baru di Indonesia Berbasis Image Processing Menggunakan Convolutional Neural Network (CNN)

Hamidah, Salma Qotrunnada (2023) Pengenalan Plat Nomor Baru di Indonesia Berbasis Image Processing Menggunakan Convolutional Neural Network (CNN). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 02311940000160-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (8MB) | Request a copy

Abstract

Seiring dengan pesatnya perkembangan teknologi maka semakin banyak teknologi yang dapat digunakan untuk membantu pekerjaan manusia. Contohnya, Sistem pendeteksian, pengenalan, dan pengidentifikasian plat nomor kendaraan. Sistem ini telah di aplikasikan di Indonesia dalam berbagai bidang, seperti pendisiplinan pengendalan terhadap lalu lintas di jalan raya dan keamanan kendaraan dalam tempat parkir. Tahapan dari sistem deteksi dan pengenalan karakter pada plat nomor kendaraan dimulai dengan pendeteksian plat nomor kendaraan, kemudian pendeteksian karakter pada plat nomor, dan prediksi karakter pada plat nomor kendaraan. Sistem yang digunakan pada penelitian ini menggunakan algoritma YOLOR dengan bantuan platform Roboflow untuk pendeteksian plat nomor dan algoritma ResNet-50 untuk pengenalan pada karakter plat nomor kendaraan. Model ResNet-50 dilakukan proses training yang kemudian diperoleh nilai akurasi 92.1% dan proses testing diperoleh nilai akurasi 90%. Nilai akurasi tersebut menunjukkan hasil yang baik, keadaan tersebut dibuktikan dengan tingkat akurasi training dan testing yang didapatkan melebihi angka 85%. Untuk mengetahui performasi sistem pengenalan plat nomor kendaraan, dilakukan pengujian sistem. Dari pengujian sistem, didapat persentase keberhasilan dalam mendeteksi plat nomor sebesar 97.91%, keberhasilan dalam mengenali karakter sebesar 70.1%, dan persentase pembacaan karakter plat nomor yaitu sebesar 25.53%.
========================================================================================================================================
Along with the rapid development of technology, there are technologies that can be used to help human work. For example, a system for detecting, recognizing, and identifying vehicle license plates. This system has been applied in Indonesia in various fields, such as disciplining control of traffic on the highway and vehicle security in parking lots. The stages of the character detection and recognition system on the vehicle license plate begin with the detection of the vehicle license plate, then the detection of characters on the license plate, and the prediction of characters on the vehicle license plate. The system used in this research uses the YOLOR algorithm with the help of the Roboflow platform for license plate detection and the ResNet-50 algorithm for recognition of vehicle license plate characters. The ResNet-50 model is carried out in the training process which then obtained an accuracy value of 92.1% and the testing process obtained an accuracy value of 90%. The accuracy value shows good results, this situation is evidenced by the level of training and testing accuracy obtained exceeding 85%. To find out the performance of the vehicle license plate recognition system, system testing is carried out. From system testing, the percentage of success in detecting license plates is 97.91%, the success in recognizing characters is 70.1%, and the percentage of reading license plate characters is 25.53%.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Convolutional Neural Network, Image Processing, New License Plate, Recognition, Convolutional Neural Network, Image Processing, Pengenalan, Plat Nomor Baru
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Industrial Technology and Systems Engineering (INDSYS) > Physics Engineering > 30201-(S1) Undergraduate Thesis
Depositing User:	Salma Qotrunnada Hamidah
Date Deposited:	08 Aug 2023 07:39
Last Modified:	08 Aug 2023 07:39
URI:	http://repository.its.ac.id/id/eprint/104186

Actions (login required)

View Item