Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Eyes on the Streets: Leveraging Street-Level Imaging to Model Urban Crime Dynamics

Zhixuan Qi, Huaiying Luo, Chen Chi

TL;DR

The paper investigates how the built environment, captured via street-level imagery, relates to urban crime dynamics in New York City. It combines semantic segmentation (DeepLabV3) to extract visual street features with multiple regression models to predict community crime rates, using 71 communities as the analytical unit. A key finding is that proximity to airports, as captured by an Aeroplane feature, correlates with higher crime rates and that gradient boosting methods outperform other regressors under the studied conditions. The work acknowledges limitations of crime-rate proxies and proposes future work to incorporate richer environmental variables and finer geographic granularity (e.g., census blocks) to enhance predictive reliability and policy relevance.

Abstract

This study addresses the challenge of urban safety in New York City by examining the relationship between the built environment and crime rates using machine learning and a comprehensive dataset of street view images. We aim to identify how urban landscapes correlate with crime statistics, focusing on the characteristics of street views and their association with crime rates. The findings offer insights for urban planning and crime prevention, highlighting the potential of environmental design in enhancing public safety.

Eyes on the Streets: Leveraging Street-Level Imaging to Model Urban Crime Dynamics

TL;DR

The paper investigates how the built environment, captured via street-level imagery, relates to urban crime dynamics in New York City. It combines semantic segmentation (DeepLabV3) to extract visual street features with multiple regression models to predict community crime rates, using 71 communities as the analytical unit. A key finding is that proximity to airports, as captured by an Aeroplane feature, correlates with higher crime rates and that gradient boosting methods outperform other regressors under the studied conditions. The work acknowledges limitations of crime-rate proxies and proposes future work to incorporate richer environmental variables and finer geographic granularity (e.g., census blocks) to enhance predictive reliability and policy relevance.

Abstract

This study addresses the challenge of urban safety in New York City by examining the relationship between the built environment and crime rates using machine learning and a comprehensive dataset of street view images. We aim to identify how urban landscapes correlate with crime statistics, focusing on the characteristics of street views and their association with crime rates. The findings offer insights for urban planning and crime prevention, highlighting the potential of environmental design in enhancing public safety.
Paper Structure (34 sections, 3 equations, 5 figures, 4 tables)

This paper contains 34 sections, 3 equations, 5 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Datasets
  3. Street View
  4. Crime Rate
  5. Geographical Units
  6. Model
  7. Semantic Segmentation
  8. Machine Learning in Crime Prediction
  9. Method
  10. Semantic Segmentation
  11. Regression Models
  12. Linear Regression
  13. Random Forest
  14. Decision Tree
  15. XGBoost
  16. ...and 19 more sections

Figures (5)

  • Figure 1: Sample points from the network.
  • Figure 2: Mean Square Error in respect of $\epsilon$
  • Figure 3: Random Forest Regression
  • Figure 4: Decision Tree Regression
  • Figure 5: Gradient Boosting Regression