Safe roads and road networks are of great importance in our society, as they help us safely and quickly get to where we need to go. Designing as well as evaluating these designs, however, is not a trivial task. Prior research has explored the classification of traffic accidents, but analysing the danger levels of road networks by means of image classification has not been done yet, at least not to our knowledge. In this paper, Killian RIjnbergen and I presented our take on a machine learning (image recognition) model that can classify maps of our road network based on the number of accidents that occur at a certain location. This way, we can provide road designers and city planners with a way to evaluate their design before the implementation has even started.