SMART DETECTION SYSTEM FOR UNAUTHORIZED PARKING IN DISABILITY SPACES AT A UNIVERSITY: DESIGN, IMPLEMENTATION, AND PROTOTYPE VALIDATION

Unauthorized occupation of parking spaces reserved for people with disabilities creates accessibility difficulties and increases the workload required for manual parking enforcement. This paper presents the design, implementation, and prototype validation of a low-cost smart detection system developed for disability parking spaces at King Abdulaziz University. The proposed system combines a Raspberry Pi 4, an Arducam IMX477 camera, a mobile permit-declaration application, English license-plate recognition, a SQLite database, a Flask-based administrative dashboard, evidence-image processing, and automated email notification. An authorized user scans a permit barcode through the mobile application and creates an active parking declaration. When a vehicle enters the monitored parking space, the Raspberry Pi detects its presence, confirms that it remains stationary for approximately 5 s, recognizes the English portion of the Saudi license plate using an OCR-based pipeline, and compares the result with the plate associated with the active declaration. Events are classified as authorized access, no-declaration violation, plate-mismatch violation, or technical detection issue. Prototype validation included vehicle-presence detection, stop-confirmation, authorization and violation classification, evidence quality and privacy processing, alert transmission, and continuous power and reliability testing. All six controlled vehicle-presence cases and all six stop-confirmation scenarios produced the expected behavior. Four authorization and violation test cases were also classified correctly. Evidence images were captured at 1920 × 1080 resolution, while four email-alert trials achieved successful delivery with an observed delay of 11 s per trial. During a two-hour continuous-operation test, the system consumed approximately 7.25 W on average and 10.25 W at estimated peak load, with an average Raspberry Pi temperature of 33.1°C and no observed under-voltage or thermal-throttling events. License-plate recognition was functional under controlled conditions but remained sensitive to lighting, distance, viewing angle, glare, blur, and plate visibility. The results demonstrate the feasibility of a low-cost edge-based system for supporting accessibility-parking enforcement while identifying license-plate recognition and long-term outdoor operation as the principal areas requiring further development.