<![CDATA[Objective: Campus drop-off areas often experience congestion due to vehicles stopping beyond the permitted duration without effective monitoring. This research aims to develop a real-time system for detecting and calculating vehicle stopping duration in campus drop-off areas using CCTV cameras and the YOLOv8 deep learning model. Method: This research applies a research and development (R&D) approach with a case study at the drop-off area of the Electrical Engineering Building, Campus 2, Ujung Pandang State Polytechnic (PNUP). The dataset, managed through Roboflow, comprises 803 images of cars and motorcycles (640×640 pixels), split into 78% training, 20% validation, and 2% test data. Model training and testing were conducted in Google Colab, while Visual Studio Code served as the main code editor. Results: The YOLOv8 model at epoch 42 of 50 achieved precision 90.8%, recall 98.5%, mAP50 94.5%, and mAP50-95 73.3%. The system successfully detects vehicles in real-time, recognizes drop-off boundaries, calculates stopping duration, and delivers voice notifications via speaker along with violation reports to a website dashboard and Telegram Bot. Novelty: This research integrates YOLOv8 object detection with polygon-based boundary recognition, automatic duration calculation, voice alerts, and dual monitoring through a website dashboard and Telegram Bot. The integration offers an automated solution for managing campus drop-off areas, reducing congestion and improving traffic efficiency.]]>
Copyrights © 2026
