Are Safer Looking Neighborhoods More Lively?

Abstract

Policy makers, urban planners, architects, sociologists, and economists are interested in creating urban areas that are both lively and safe. But are the safety and liveliness of neighborhoods independent characteristics? Or are they just two sides of the same coin? In a world where people avoid unsafe looking places, neighborhoods that look unsafe will be less lively, and will fail to harness the natural surveillance of human activity. But in a world where the preference for safe looking neighborhoods is small, the connection between the perception of safety and liveliness will be either weak or nonexistent. In this paper we explore the connection between the levels of activity and the perception of safety of neighborhoods in two major Italian cities by combining mobile phone data (as a proxy for activity or liveliness) with scores of perceived safety estimated using a Convolutional Neural Network trained on a dataset of Google Street View images scored using a crowdsourced visual perception survey. We find that: (i) safer looking neighborhoods are more active than what is expected from their population density, employee density, and distance to the city centre; and (ii) that the correlation between appearance of safety and activity is positive, strong, and significant, for females and people over 50, but negative for people under 30, suggesting that the behavioral impact of perception depends on the demographic of the population. Finally, we use occlusion techniques to identify the urban features that contribute to the appearance of safety, finding that greenery and street facing windows contribute to a positive appearance of safety (in agreement with Oscar Newman's defensible space theory). These results '16, October 15 -19, 2016, Amsterdam, Netherlands suggest that urban appearance modulates levels of human activity and, consequently, a neighborhood's rate of natural surveillance. MM

The amenity mix of urban neighborhoods

“…(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.) amenity mix of a neighborhood can be explained by place characteristics that are not included in our model, such as architectural or historic value (Been et al, 2016;De Nadai et al, 2016;Naik et al, 2017Naik et al, , 2016Salesses et al, 2013) or environmental externalities of car use. The lesson here is that the model successfully detects a known reality of Harvard Square, which is a lack of parking relative to the number of amenities it hosts.…”

Using Convolutional Networks and Satellite Imagery to Identify Patterns in Urban Environments at a Large Scale

“…Urban environments have been analyzed using other types of imagery data that have become recently available. In [4,14], the authors propose to use the same type of imagery from Google Street View to measure the relationship between urban appearance and quality of life measures such as perceived safety. For this, they hand-cra standard image features widely used in the computer vision community, and train a shallow machine learning classi er (a support vector machine).…”

Systematic Social Observation at Scale: Using Crowdsourcing and Computer Vision to Measure Visible Neighborhood Conditions

“…Naik et al (2014) developed an algorithm called Streetscore to predict the perceived safety of more than 1 million images from 21 U.S. cities. These data have been used to better understand predictors of urban perception (Porzi et al 2015), identify the impact of historic preservation districts on urban appearance (Been et al 2016), determine the effects of urban design on perceived safety (Harvey and Aultman-Hall 2016), compare liveliness with mobile phone data (De Nadai et al 2016), and examine changes over time (Naik et al 2017). Although informative, these developments in the field of CV are still limited because they are based on perceptions from anonymous respondents.…”