Measuring the Built Environment with Google Street View and Machine Learning: Consequences for Crime on Street Segments

Abstract

Objectives: Despite theoretical interest in how dimensions of the built environment can help explain the location of crime in micro-geographic units, measuring this is difficult.Methods: This study adopts a strategy that first scrapes images from Google Street View every 20 meters in every street segment in the city of Santa Ana, CA, and then uses machine learning to detect features of the environment. We capture eleven different features across four main dimensions, and demonstrate that their relative presence across street segments considerably increases the explanatory power of models of five different Part 1 crimes.Results: The auto-oriented measures-vehicles and pavement-were positively associated with crime rates, as was the presence of more persons in the environment. For the defensible space measures, the presence of walls has a slowing negative relationship with most crime types, whereas fences did not. And for our two greenspace measures, although terrain was positively associated with crime rates, vegetation exhibited an inverted-U relationship with two crime types.Conclusions: The results demonstrate the efficacy of this approach for measuring the built environment.

Modeling the Effect of Streetscape Environment on Crime Using Street View Images and Interpretable Machine-Learning Technique

“…These theories emphasize the importance of the built environment in crime research from different perspectives. The rational choice theory argues that certain built-in environmental features can influence offenders’ behaviors by affecting their perceptions of risk and cost [ 1 , 2 ], which in turn affects the spatial distributions of crime. The routine activity theory claims that the built-in environmental features can impact the spatial-temporal overlap of offenders, potential victims, and ineffective supervisors, which in turn affects the occurrence of crime [ 3 ].…”

“…Walls and fences could control the accessibility of the area, increasing the difficulty of escaping after committing crimes. Therefore, the existence of walls and fences could somewhat help inhibit the occurrence of crime [ 1 , 16 , 17 ]. The appearance of street lights [ 18 ] or surveillance cameras [ 19 ] reflects the surveillance level of a territory.…”

“…Green elements such as trees and grass can facilitate people’s outdoor activities and thus promote informal monitoring, thereby restraining the occurrence of crimes. However, thick shrubs may obstruct people’s lines of sight and provide hiding places for offenders; therefore, areas with shrubs have potential crime risks [ 1 , 23 , 24 , 25 ]. A holistic investigation of these features and their possible interactions remains unanswered.…”

“…As computer vision technology matured, researchers gradually combined deep learning image extraction methods with street view images to extract large-scale street microenvironment features. For example, Hipp [ 1 ] used deep learning methods to extract pixel proportions of streetscape elements such as buildings, people, sidewalks, vehicles, fences, and walls from Google Street View images. Based on this data, they analyzed the relationship between different dimensions of environmental characteristics and different types of crime.…”

Identifying Streetscape Features Using VHR Imagery and Deep Learning Applications

“…Given the broad coverage of these street views APIs, these tools have proved beneficial for analyzing multiple cities at once [15]. More specifically, these methods have been utilized to find the relationships between the visual appearance of streetscapes and the health of the citizens [16,17], safety and crime [18,19], and urban aesthetics [20][21][22]. Further, such studies Similarly, identifying streets from satellite and aerial images has gathered a lot of interest in recent years.…”