No image available for this title

Skripsi

Comparative analysis tensorflow and yolov8 for automatic number plate recognition based vehicle detection with easyocr integration / SALAH ABDULLAH KHALIL ABDULRAHMAN

Abstrak
Automatic Number Plate Recognition (ANPR) systems play a vital role in various applications including law enforcement parking management and toll collection. In this study a comparative analysis between TensorFlow and YOLOv8 models is conducted for ANPR-based vehicle detection with integration of the EasyOCR library for optical character recognition. This study conducts a Comparative Analysis of TensorFlow SSD MobileNetV2 and YOLOv8 for Automatic Number Plate Recognition (ANPR) Based Vehicle Detection with EasyOCR Integration addressing a significant gap in existing research. While previous studies focused on static image testing this research explores their effectiveness in real-time video-based scenarios crucial for dynamic environments. Employing YOLOv8m and TensorFlow 2.10 with SSD MobileNetV2 the study aims to compare accuracy precision recall and F1-Score metrics. With a focus on real-world applicability and integration with EasyOCR this research aims to inform the development of more efficient ANPR systems for diverse scenarios enhancing reliability and usability. In implementing the YOLOv8m and TensorFlow with SSD MobileNetV2 models the AI Project Life Cycle method was meticulously followed to ensure a systematic approach from problem identification to deployment. Both models underwent testing using the same parameters including the number of epochs optimizer function and batch size to facilitate a fair comparative analysis. The evaluation focused on videos prepared beforehand utilizing a Windows device with NVIDIA GeForce RTX 3070 Ti 8GB. Accuracy Precision Recall and F1 Score metric values were derived from the confusion matrix providing a detailed breakdown of the model s predictions compared to the actual ground truth. SSD MobileNetV2 exhibited moderate performance metrics with an accuracy of 62% vii precision of 82.3% recall of 68.3% and an F1 score of 75.6%. However YOLOv8 outperformed SSD MobileNetV2 demonstrating superior accuracy of 92.8% precision of 92.5% recall of 100% and F1 score of 96.1% with faster processing speeds in license plate detection. while the comparative analysis of license plate recognition models reveals compelling insights into their performance metrics. SSD MobileNetV2 achieved an accuracy of 70% precision of 70% recall of 100% and an F1 score of 82.3% while YOLOv8m exhibited higher performance with an accuracy of 80% precision of 77% recall of 100% and an F1 score of 87%. These results further emphasize YOLOv8 s efficiency and potential for real-world deployment in ANPR systems.

Informasi Detail
DDC
SKRIPSI DIGITAL
Prodi
Universitas Negeri Malang. Program Studi Teknik Informatika, .
Deskripsi Fisik
xiv, 58 hlm. : ilus.
Bahasa
No Reg
4311/RS/24
Edisi
Skripsi (Sarjana)--Universitas Negeri Malang. 2024
Subjek
1. ANALISIS PERBANDINGAN TENSORFLOW DAN YOLOV8
2. SISTEM DETEKSI PLAT KENDARAAN SEPEDA MOTOR - INTEGRASI EASYOCR
3. COMPARATIVE ANALYSIS OF TENSORFLOW AND YOLOV8

Pembimbing
1. Kartika Candra Kirana, S.pd., M.kom
Lampiran Berkas