A web platform for crowdsourced collection, processing, and visualization of exposure data on buildings

The ultimate objective of our research is to explore the potential of Machine Learning in the dynamic creation of up-to-date exposure layers for buildings. This effort involves integrating remote sensing images, ancillary data such as national census information, and crowdsourced data collected by trained citizens. The crowdsourcing activity builds on a previous successful initiative developed within the CEDAS (building CEnsus for seismic Damage Assessment) project, which engaged high school students from North-East Italy in collecting data on buildings that were either unavailable from conventional exposure data sources or not easily retrievable via remote sensing techniques (Scaini et al., 2022).

To this end, we are developing a complex multimedia information system via web platform designed to collect, process, store, and distribute information to different knowledge users (policymakers, territorial planners, citizens) with targeted visualization strategies. The crowdsourcing initiatives are taking place in selected municipalities of Tuscany and Friuli regions (Italy), exposed to different natural hazards, such as earthquakes, tsunamis and landslides.  An online questionnaire has been created to assist the user in building data collection and minimize input errors. Simultaneously, building data, along with their photos, are stored in a structured database for research purposes.  For instance, building data and images are used as learning set to train a machine learning algorithm to identify specific features such as roof type, number of floors, and the presence of a basement. These algorithms can then be included in the online questionnaire to facilitate further data collection by automatically suggesting features associated to the buildings. A dedicated visualization tool is being developed on the web platform to showcase the effectiveness of this method in recognition of building features. We will demonstrate the data visualization tools developed on the web platform so far, highlighting the key features of the available exposure databases. The web platform is designed to provide an easy-to-use tool for communicating with various knowledge users, while also enhancing disaster awareness and preparedness, which is attained exploring and collecting data on the built environment.

This study is a contribution to the ongoing PRIN 2022 PNRR project SMILE “Statistical Machine Learning for Exposure development” (code P202247PK9, CUP B53D23029430001) within the European Union-NextGenerationEU program.