AI, built environment and patterns of safety perceptions in Stockholm
Stockholm is today facing a number of challenges, with poor declared safety levels being a prominent one. Different groups of people may feel at more risk than others, with some even actively avoiding public places due to their fear. The question that arise is: which characteristics of the urban environment are perceived as safe or unsafe by the population?
The aim of the study is to determine how safety perceptions relates to the urban and social landscape of Stockholm. This is achieved by creating an AI safety perception map based on several different data sources. Data from online safety surveys will be combined together with street view imagery using AI, which later will be compared with multiple urban security indicators (e.g. police records, calls for street services and disorder). The end goal is to develop a prototype of a long-term monitoring platform using AI that is reproducible, providing planners with a knowledgebase of how safety patterns can vary temporally and by groups of population, as well as which urban features are most commonly perceived as (un)safe.
Project leader (KTH): Vania Ceccato in collaboration with Jonatan Abraham
Project leader (MIT): Fábio Duarte
The project is a part of Senseable Stockholm Lab, a collaboration between KTH Royal Institute of Technology, MIT Massachusetts Institute of Technology, and the City of Stockholm with support from the Stockholm Chamber of Commerce and Newsec.