Automatic License Plate Recognition (ALPR) systems have become essential for modern traffic monitoring and vehicle control. However, many existing ALPR solutions rely on restrictive infrastructure or high-performance hardware, making them unsuitable for open environments with limited computational resources. This paper presents a real-time ALPR system tailored for university settings, designed to operate effectively under adverse weather conditions and without physical barriers. Our approach combines traditional computer vision techniques, such as background subtraction and morphological filtering, with a fine-tuned YOLOv11-m model for accurate license plate detection. To maintain real-time performance on a mid-range CPU, we implemented a multiprocessing pipeline using queues to decouple and parallelize tasks, including frame acquisition, vehicle detection, license plate recognition, and optical character recognition (OCR). The system was deployed in a 12-hour real-time test using live-stream video. It successfully demonstrated low latency and robust operation under varying lighting and weather conditions. This work presents a practical and efficient alternative to ALPR for institutions that require flexible and cost-effective vehicle monitoring solutions.