With cities reinforcing greener ways of urban mobility, encouraging urban cycling helps to reduce the number of motorized vehicles on the streets. However, that also leads to a significant increase in the number of bicycles in urban areas, making the planning of the cycling infrastructure an important task. To measure bicycle-related traffic, actively generated data from bicycle counting stations and parked bicycles at public transport hotspots, as well as data of bicycle rental companies are typically taken into account. We want to extend this bicycle-related dataset by using volunteered social media data. Can we gain insight into spatial-temporal patterns of bicycle usage in the city by using crowdsourced volunteered geographic information?
Our approach consists of three general phases. Phase I refers to processing posts that include geotagged images. We use a subset of the YFCC100m dataset, a collection of posts from the social media platform Flickr, and utilize a state-of-the-art object detection algorithm to detect and classify moving and parked bicycles. Phase II refers to visualizing the object detection results in order to show spatial and spatio-temporal patterns of bicycle locations. We also compare the Flickr data with other datasets to evaluate whether we can provide new information. Phase III shall allow an interactive exploration of data with maps and dashboards that provide details on demand and facilitate pattern detection. With this research, we aim to support decision-making related to bicycle traffic in urban planning.
- Knura, M., Kluger, F., Zahtila, M., Schiewe, J., Rosenhahn, B., & Burghardt, D. (2021). Using Object Detection on Social Media Images for Urban Bicycle Infrastructure Planning: A Case Study of Dresden. ISPRS International Journal of Geo-Information, 10(11), 733. DOI: 10.3390/ijgi10110733