Analyzing Urban Crime Through Street View Imagery: Insights from Urban Micro Built Environment and Perceptions

Understanding the relationship between urban crime and the built environment is crucial for developing effective crime prevention strategies, particularly in the context of rapid urban development and city planning. As cities grow, urbanization leads to environments that either promote or inhibit criminal activity, making it essential to explore the interactions between urban design and crime. This study investigates the impact of micro built environment (MBE) elements and place perceptions on crime occurrences in Toronto using street view imagery (SVI) data and machine learning models. We used logistic regression models and an XGBoost (Version 1.7.5) classifier to assess the significance of MBE and perception variables in classifying crime and non-crime intersections. Our findings reveal that intersections with criminal activity tend to be related to more mobility-related features, such as roads and vehicles, and fewer natural elements, such as vegetation. The “beautiful” and “depressing” perceptions emerged as the most significant variables in explaining crime events, surpassing the commonly studied “safety” perception. The XGBoost model achieved 86% accuracy, indicating that MBE and perception variables are strong predictors of crime risk. These findings suggest that enhancing vegetation and improving street aesthetics could serve as effective crime prevention measures in urban environments. However, limitations include the general nature of the perception model and the reliance on aggregated crime data. Future research should incorporate local perceptions and fine-scale crime data to provide more tailored insights for urban planning and crime prevention