Vision based parking system faces the challenges of multi scale information gathering, contextual event detection and
the deployment of large systems (Hampapur et al. 2005). System complexity is related to the hardware and software
part. Hardware part consist of several camera sensors, network and server infrastructure. Several servers are needed for
image processing, data archiving and information dissemination. The majority of computational power is used for
image processing and high level data extraction (vehicle and incident situations recognition, parking lot occupancy
computation, license plate recognition, etc.). This presents a drawback of such a system which can be overcome by the
usage of parallel and distributed computing. Unlike classical parking lot management systems, computer vision system
generates large amounts of various data. Such data are in the form of raw video footage or extracted high level data.
Cloud based data storage presents a very efficient solution for storing significant amount of video data.