Deep mapping gentrification in a large Canadian city using deep learning and Google Street View

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

Gentrification is multidimensional and complex, but there is general agreement that visible changes to neighbourhoods are a clear manifestation of the process. Recent advances in computer vision and deep learning provide a unique opportunity to support automated mapping or ‘deep mapping’ of perceptual environmental attributes. We present a Siamese convolutional neural network (SCNN) that automatically detects gentrification-like visual changes in temporal sequences of Google Street View (GSV) images. Our SCNN achieves 95.6% test accuracy and is subsequently applied to GSV sequences at 86110 individual properties over a 9-year period in Ottawa, Canada. We use Kernel Density Estimation (KDE) to produce maps that illustrate where the spatial concentration of visual property improvements was highest within the study area at different times from 2007–2016. We find strong concordance between the mapped SCNN results and the spatial distribution of building permits in the City of Ottawa from 2011 to 2016. Our mapped results confirm those urban areas that are known to be undergoing gentrification as well as revealing areas undergoing gentrification that were previously unknown. Our approach differs from previous works because we examine the atomic unit of gentrification, namely, the individual property, for visual property improvements over time and we rely on KDE to describe regions of high spatial intensity that are indicative of gentrification processes.

Related collections

Most cited references 50

Record: found
Abstract: not found
Conference Proceedings: not found

CNN Features Off-the-Shelf: An Astounding Baseline for Recognition

Ali Sharif Razavian, Hossein Azizpour, Josephine Sullivan … (2014)

0 comments Cited 476 times – based on 0 reviews

Bookmark

Record: found
Abstract: found
Article: not found

Using Google Street View to audit neighborhood environments.

Julien O Teitler, Kathryn Neckerman, Michael Bader … (2010)

Research indicates that neighborhood environment characteristics such as physical disorder influence health and health behavior. In-person audit of neighborhood environments is costly and time-consuming. Google Street View may allow auditing of neighborhood environments more easily and at lower cost, but little is known about the feasibility of such data collection. To assess the feasibility of using Google Street View to audit neighborhood environments. This study compared neighborhood measurements coded in 2008 using Street View with neighborhood audit data collected in 2007. The sample included 37 block faces in high-walkability neighborhoods in New York City. Field audit and Street View data were collected for 143 items associated with seven neighborhood environment constructions: aesthetics, physical disorder, pedestrian safety, motorized traffic and parking, infrastructure for active travel, sidewalk amenities, and social and commercial activity. To measure concordance between field audit and Street View data, percentage agreement was used for categoric measures and Spearman rank-order correlations were used for continuous measures. The analyses, conducted in 2009, found high levels of concordance (≥80% agreement or ≥0.60 Spearman rank-order correlation) for 54.3% of the items. Measures of pedestrian safety, motorized traffic and parking, and infrastructure for active travel had relatively high levels of concordance, whereas measures of physical disorder had low levels. Features that are small or that typically exhibit temporal variability had lower levels of concordance. This exploratory study indicates that Google Street View can be used to audit neighborhood environments. Copyright © 2011 American Journal of Preventive Medicine. Published by Elsevier Inc. All rights reserved.