Table of Contents
Urban Sci., Volume 1, Issue 2 (June 2017)
- Issues are regarded as officially published after their release is announced to the table of contents alert mailing list.
- You may sign up for e-mail alerts to receive table of contents of newly released issues.
- PDF is the official format for papers published in both, html and pdf forms. To view the papers in pdf format, click on the "PDF Full-text" link, and use the free Adobe Reader to open them.
Cover Story The cover image illustrates urban mapping through street-level images as implemented by Ticinum [...] Read more. The cover image illustrates urban mapping through street-level images as implemented by Ticinum Aerospace. Starting from a public map of the area of interest, such as an OpenStreetMap layer, the system defines a georeferenced “visit path”, along which street-side building pictures are sought for, and harvested from, publicly accessible repositories. The retrieved georeferenced images are fed into an opportunely trained deep neural network. This latter consequently finds determining features and labels the building according to a given taxonomy, in addition to determining specific parameters such as floor count, resulting in a remarkably enriched GIS layer being output. This paper shows how deep learning enables leveraging on the wealth of available crowdsourced pictures to benefit practical applications including, for example, enhanced exposure models for risk assessment, or real estate valuation. View this paper.