AI Based Automated Traffic Monitoring System for Vehicles and License Plate Recognition

by Read DANJOLLI

Abstract
This paper presents a comprehensive architecture for automated traffic violation surveillance. It is based on sophisticated deep learning algorithms and artificial intelligence systems with computer vision. The main objective is to develop an integrated pipeline that integrates vehicle detection, Automatic License Plate Recognition (ALPR), and visual attribute classification (e.g., color, manufacturer, and model). YOLO detection, DeepSORT tracking, CRNN network OCR, and CNN for car brand and color categorization are all parts of the technical solution. The study fully compares Edge and Cloud architectures, examining how well they perform under different conditions, such as high traffic and poor lighting. The findings show that, while Cloud solutions offer more flexibility but at a higher latency cost, Edge solutions, despite their processing limitations, achieve response times below 200 ms and accuracy above 95% in license plate identification.
Along with specific implementation recommendations for the Albanian context, the study addresses algorithmic fairness, privacy protection and GDPR compliance. It also addresses the ethical and legal elements of using surveillance technologies, highlighting the prospects and challenges for a successful adoption in Albania.
Furthermore, to compensate for the personalized data pages for the Albanian market, synthetic data models were included in the initial training. This was sufficient on the ground to allow for higher algorithmic adaptability. Investments in human resource training and a well-defined framework are also necessary for the deployment of technologies to ensure accountability, transparency and responsibility for all. This comprehensive strategy lays the foundation for an automated application system in Albania that is reliable and sustainable.

Key words: Automated Traffic Surveillance, Automatic License Plate Recognition, Vehicle Attribute Recognition, Deep Learning & Computer Vision, Edge and Cloud Architecture, Ethical & GDPR Compliance