Albeit many decisions are being made to improve cycling facilities, both in improving the infrastructure and providing educational benefits of cycling, increasing the number of cyclists have proven difficult. In other words, mechanisms must be carried out such that prospective cyclists are encouraged to adopt cycling. However, the current cycling panorama prevents most potential cyclists from beginning to cycle. For one, although cycling fatalities have decreased in Europe in the past years, in 2016 it still represented almost 8% of all fatalities in roads (European Comission, 2018). Cyclers face many adversities in their daily journeys, whether it is interactions with cars or ill-designed infrastructure and, as a result, these interactions affect potential cyclists’ decisions to keep on cycling or to look for an alternative mode of transportation. In this sense, studying perceived risk and how different factors influence the perception of safety is key. Current approaches are not scalable and consist mainly in performing interviews or deploying surveys.

With this in mind, we propose our research to interlink data sensing, computer vision, image processing and signal processing to create an advantage over other research that has been carried on the topic of cyclists’ perception of safety. By applying machine learning and image processing methods we might be able to scale inferences of how humans perceive risk.