the other street view

 

Computational strategies for finding patterns of inequity in Google Street View Images.
Equity, Machine Learning, Image Mining, Pattern Recognition, Social Justice, Google Street View, City Data, Urban Inequality

sorting Street View images with various attributes

sorting Street View images with various attributes

Sorting Google Street View Images

The Google Street View project was ultimately not about collecting images of the street, but about collecting data from the images of the street. At the Street View project's onset there was an understanding that photos at the street level held an immense amount of information, however it wasn't yet clear how all of that information would be used. Currently it is difficult to use them for much more than navigation because photos are organized geographically by their latitude and longitude location. This thus limits major analysis of street photos. For example, its currently not possible on Google's platform to sort and look at Street View images of all prison locations in the United States, the most expensive houses in a city, or find street images at places with bad air quality. In order to analyze the immense archive of street view photos beyond Google's interface of tapping on arrows to glide up and down streets it is necessary to hack Google's API to pull images by location.

“If Google's mission is to organize all the world's information, the most important challenge far larger than indexing the web is to take the world's physical information and make it accessible and useful.”

Inequity In the Built Environment

When browsing down streets in Google the differences between neighborhoods is plain as day. People live in completely different worlds. In one world homes are hidden behind power lines, dollar stores are painted with graffiti, and police park around every corner waiting for suspects. A few miles away another world seems completely stranger. Well-kept brownstones are lined with trees and potted plants. Starbucks with glass storefronts seem to insist on opportunity for productive creativity at the purchase of a coffee. While its naive to suggest that neighborhoods should mimic a polished wealthy neighborhood, there is no denying that a person growing up in the second neighborhood will likely have the privilege and access to more opportunities than someone in the first neighborhood. The intent of this project is to uncover cyclical patterns of investment and disinvestment in neighborhoods by comparing the street view data set at low and high income neighborhoods.

If urban form affects social outcome, street view images can uncover patterns in the urban fabric and be used to analyze the success or failures of urban form.

Money for Neighborhoods Not Jails

Prejudiced policies about safety in the urban environment have lead to over-policing in low-income neighborhoods. Theories for Urban safety have spanned a wide range historically from Jane Jacob's "Eyes On the Street" to the highly controversial Broken Windows Theory. The issue with policies like Broken Windows is that when a window is broken, it is not an indication that residents should be policed or penalized, but rather an indication that the residents lack economic capability to fix the window. Rather than criminalize poverty and continue to spend money on policing and jailing, investment should be placed at the root of the problem. Making better neighborhoods, and producing economic opportunities in neighborhoods that need it. 

Classifying urban form can’t be dismissed as solely aesthetic! Our urban environment impacts our perception of ourselves in society, whether we feel empowered, respected, whether we feel like a part of our neighborhood or like an outsider. 

Cyclical patterns begin to develop in both neighborhoods. When a glassy cafe moves into one neighborhood, higher-income residents move in, developers invest money in greening the streets, cleaners move in to service higher-income residents who have surplus money for their chores, landlords make signage gold and flashy to prove how luxurious their building is.

 
vacant lot location in The Upper West Side (left) and Brownsville (right)

vacant lot location in The Upper West Side (left) and Brownsville (right)

 

Brownsville and The Upper West Side

If urban form affects social outcome, what characteristics of a neighborhood’s urban form predispose its residents to greater levels of policing and incarceration? This project aims to make visible the differences in the urban fabric by using the Google Street View API with coordinates from a database of vacant lot locations between 2 neighborhoods: Brownsville, a low-income neighborhood in Brooklyn and the Upper West Side, a high-income neighborhood in Manhattan. Brownsville and the Upper West Side were chosen because they have similar population size and are similar in size. However they are very different economically and policed differently. 

Looking at vacant lot locations in Brownsville and the Upper West Side.

Issue with vacant lot database - ownership is incorrect

Sorting Vacant Lot Locations By Stop and Frisk Amounts

very different worlds - one of brownstones line with trees and potted plants, Starbucks with glass storefronts, and clean streets, versus homes behind powerlines, dollar stores painted with graffiti, and police parked around the corner

Both locations become self-fulfilling prophecies - driving towards extreme environments of wealth versus poverty

strategies for finding patterns of disinvestment by comparing large data sets of street view images in high and low-income neighborhoods
 
 

Machine Learning Image Analysis

machine learning

 
scripts used

scripts used

scrambled Latin used by designers to mimic real copy. Donec eu est non lacus lacinia semper. Sed a ligula quis sapien lacinia egestas. Integer tempus, elit in laoreet posuere, lectus neque blandit dui, et placerat urna diam mattis orci. Mauris egestas at nibh nec finibus. Vivamus sit amet semper lacus, in mollis libero.

ec ac fringilla turpis. Class aptent taciti sociosqu ad litora torquent per conubia nostra, per inceptos himenaeos. Aenean eu justo sed elit dignissim aliquam.

 

ollowing is placeholder text known as “lorem ipsum,” which is scrambled Latin used by designers to mimic real copy. Sed a ligula quis sapien lacinia egestas. Donec ac fringilla turpis. Donec eu est non lacus lacinia semper. Lorem ipsum dolor sit amet, consectetur adipiscing elit.

g is placeholder text known as “lorem ipsum,” which is scrambled Latin used by designers to mimic real copy. Integer tempus, elit in laoreet posuere, lectus neque blandit dui, et placerat urna diam mattis orci. Suspendisse nec congue purus. Nulla eu pretium massa.

2.jpg
0.jpg