WinterWatch: An Automated Approach to Outdoor Parking Lot Vacancy Detection in Cold Climates

The increasing demand for efficient parking space utilisation has triggered the development of vacancy detection systems, with a particular focus on addressing the unique environmental constraints imposed by winter conditions. In this paper we present vacancy detection tool trained on our dataset that was collected in the parking lot of the University of Zilina, capturing unique winter conditions from December to January. The dataset includes diverse weather scenarios, such as snowfall or early morning conditions where light is reflected directly into the camera. Leveraging YOLOv4, Faster R-CNN, and PKLot dataset, WinterWatch demonstrates good performance in real-time vacancy detection under these challenging circumstances. WinterWatch not only contributes to the evolution of smart parking systems but also underscores the importance of tailored solutions for diverse environmental settings, ensuring year-round efficiency in outdoor parking management.