The No Parking Vehicle Detection System aims to automate the process of identifying vehicles parked in unauthorized or restricted areas using advanced computer vision and deep learning techniques. The system utilizes live streaming from surveillance cameras installed near “No Parking” zones. The captured video is divided into individual frames, which are then subjected to pre-processing operations such as noise reduction, resizing, and normalization to enhance image quality.

A MobileNet-based deep learning model is trained on a large vehicle dataset containing labeled images of cars captured in various lighting and environmental conditions. The region of interest (ROI) is defined to focus the detection process only on the area where parking violations are likely to occur. Once the model identifies an illegally parked vehicle, a decision-making module triggers a buzzer alert to notify nearby authorities and simultaneously sends a digital message to the RTO office with relevant details such as location, time, and vehicle image evidence.

This automated detection system reduces the need for manual monitoring, helps improve traffic discipline, and ensures faster enforcement of parking rules. The integration of IoT, computer vision, and machine learning makes the system efficient, cost-effective, and scalable for deployment in smart city infrastructure. Additionally, the project can be expanded by incorporating license plate recognition and GPS-based mapping for enhanced accuracy and reporting.

Keywords: Mobile net,Parking system,RTO,Decision making.