Video-based parking space detection

Finding a vacant parking lot in urban areas is time-consuming and, thus, not satisfying for potential visitors or customers. Efficient car-park routing systems could support drivers to get an optimal parking lot immediately. Current systems detecting vacant parking lots are either very expensive due to hardware requirement for each parking lot or do not provide a detailed occupancy map. In this paper, we propose a video-based system for low-cost vacant parking space detection. A wide-angle lens camera is used in combination with a desktop computer. Different feature extractors and machine learning algorithms were evaluated in order to retrieve accurate state information for each of the observed parking lots. We found a combination of feature extractors and classifiers which properly solved the given task. Our final system, incorporating temporal integration, reached an accuracy of 99.8 %.