Paradapp: an Intelligent Car Parking Assistant Mobile Application Using Convolutional Neural Networks for Student and Novice Drivers

Parking is a significant challenge for many drivers, often leading to stress and accidents. This paper presents Paradapp, an intelligent car parking assistant designed to aid both student and novice drivers. Utilizing Convolutional Neural Networks (CNN), specifically the YOLO object detection module, Paradapp leverages vehicle’s car cameras to detect obstacles and estimate distances in real-time. The application provides voice-prompted instructions to guide drivers through the parking process, enhancing safety and efficiency. The study aims to reduce parking-related stress and accidents, offering a smart solution accessible via mobile devices. Paradapp’s development follows the AGILE methodology, ensuring iterative improvements and user-centric design. To train the model, a total of 14,000 images were collected from the KTTI and COCO repository to form a dataset related to parking and driving scenarios. The mean average precision of the YOLOv8 model is at 74.3% while the F1 score is at 73% and the Recall score is at 87%. This research contributes to the growing field of intelligent driving systems, providing valuable insights for future advancements in driver-assistance technologies. The system’s effectiveness is evaluated based on ISO 25010:2011 software quality standards, focusing on functional suitability, performance efficiency, usability, reliability, and safety. The overall acceptance of the system is 4.49, interpreted as “Agree”.