Making Street View Sortable
“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.” (1)
At the onset of Google Street View the project team understood that street level photos held an immense amount of information. However, it was not yet a clear how all that information would be useful. In Google Maps, images of the street are sorted geographically, which limits its usefulness to exploring street images one-by-one. Finding out what all of the schools in New York look like requires panning across Google Maps to each location and dropping the peg man in front of each school- not exactly ideal especially for researchers trying to draw comparisons between thousands of images. Google doesn’t yet have an accessible way to sort street view by other information in the data, i.e. you cannot view all street view prison locations in the US, or look at all places with the highest stop and frisk in New York.
The following method shows how to collect street view images in mass by using a list of the desired dataset’s addresses. For this example I used Family Dollar Stores locations because they are quickly recognizable in images and their locations are readily available online. The Project begun by scraping a list of store locations from the Family Dollar website. The data was downloaded as a csv file and then cleaned in excel to separate addresses into a single column.
The addresses were put into a Batch Geocoding website that translates the addresses into latitude and longitude coordinates. This list of coordinates was then used with Google’s Street View Image API. I used a Python script to pull the images from the web and batch name them. The script can be downloaded on GitHub.
The result is a folder of batch-named Google Street View images files.
