Car park occupancy analysis using UAV images

With the development of unmanned aerial vehicles (UAVs) and the relevant techniques, UAVs become common and popular for civilian applications such as remote sensing tasks. The reason is because they are cheap, flexible, and easy to set up. Car park occupancy analysis is important for authorities to make decisions on the design, plan and management of car parks. To have a quick knowledge of current parking situations, we proposed to use UAV images to count how many cars are parked during different periods. In this paper, our major contribution is a novel car counting approach for UAV images. Different from traditional detection- or segmentation-based counting techniques, the proposed counting method is density estimation based that does not need intense collection and learning procedures. We transform the car counting problem into the estimation of density values over pixels of an image. Experimental results have been conducted on real car park scenarios and all the results show that our method can provide a promising estimation of car numbers.