Image-based Analysis of Aerial Imagery for Crowd Counting Application

Bachelor Thesis, Calvin Kramer, 2020

Due to ongoing population growth and the resulting increase in cities’ densities, we are facing a world in which blank space becomes less common and crowded areas the everyday standard of many people’s life. Although foot traffic is a well-known size in urban planning, bottlenecks cannot always be avoided. Especially in places where people tend to congregate e.g. shopping malls, festivals or train stations. These bottlenecks can rapidly turn into dangerous situations causing death like the love parade disaster in Duisburg Germany on the 24th of July 2010. Therefore, the amount of people in a certain area is not only a quantity that should be considered once during the planning stage of neighborhoods or festivals but also be constantly measured so that shortage of space can be detected and handled properly. Whereas there are many approaches when it comes to crowd counting systems, the following work will exclusively deal with the approach of using deep learning algorithms and aerial imagery.