<![CDATA[This study presents the design and implementation of a real-time Automatic License Plate Recognition (ALPR) system specifically tailored for Indonesian vehicle plates, integrating EasyOCR, computer vision-based image preprocessing, and Regular Expression (regex) validation. The system captures images or video streams and applies a multi-level preprocessing pipeline, including grayscale conversion, Gaussian noise reduction, edge detection, and contour-based plate localization, before performing optical character recognition based on deep learning using a convolutional recurrent neural network with an attention mechanism. Post-recognition processing with regex filtering ensures strict compliance with the official Indonesian license plate format, thereby minimizing false positives and improving recognition accuracy. Experimental evaluation using real-world surveillance data achieved 75% accuracy, 100% precision, 75% recall, and an F1-score of 86%, indicating an optimal balance between detection precision and sensitivity. The system’s advantages include real-time performance, ease of deployment with open-source software, and adaptability to various lighting and environmental conditions. However, the system still shows limitations under extreme conditions such as nighttime, heavy rain, and dense traffic, where recognition accuracy tends to decrease. Therefore, future research will focus on algorithm optimization for low-light, adverse weather, and motion-blur scenarios, large-scale deployment in urban areas, and integration with AI-based vehicle tracking, positioning this system as a key enabling technology in the development of smart city infrastructure.]]>
Copyrights © 2025
