The occurrence of crime in a neighbourhood is a threat to public safety and decreases citizens’ quality of life. Modelling the relationships between environmental characteristics (physical and social) and different crimes is a key research challenge in Geography. Most existing studies have focused on single crime types (e.g., burglary or sexual assault), while it is still unclear why different crime types occur together and whether there is any mechanistic basis for the geographic patterns of crime diversity. This project aims to uncover the contribution of the physical and social environment to crime diversity, via large-scale spatial analyses of multiple crimes at the street-segment level, covering all 308 municipalities in the Flemish Region of Belgium (i.e., Flanders). Specifically, we will 1) propose an automated method to quantify the fine-grained attributes of the environment based on street-view images; 2) elucidate the multi-scale links between environment and crime diversity across Flanders. Practically, the project results could inform local communities, policymakers, and law enforcement agencies to improve crime prevention that adapts to the characteristics of the built and social environment.