Flickr Cities

Introduction

Since its start in 2004, Flickr has been one of the most popular photo-sharing platforms on the Internet. There are other widely used image sharing services but Flickr has the deepest historical record of geo-coded photographic content.

We've long been interested in static maps of social network activity in cities, and how the coordinates aggregate into discernible geographic structures whilst other spaces are left as a void. In addition to this we wanted to explore the historical dimension of urban data, to unearth to what extent different kinds of events and phenomena are recorded - a kind of archaeology of Flickr data.

In this release we have produced an interactive application that expresses the spatial configurations, temporal changes, and taxonomies of the data.

Example Map Images

We've selected a cross-section of world cities as case studies. Flickr is obviously used more widely in some countries than others, so usage isn't likely to reflect the population size of a city. We have data for the largest Megacities by population, so that it’s possible to compare the urban activity of the Flickr community in different parts of the world. We also have sets for the Americas and Europe where Flickr use is more concentrated. For the cities with large datasets we are using a years worth of data - 2012. Where possible we are looking at larger date ranges from 2005 - 2012.


Timeline

We've used the Flickr API to generate temporal city level maps. Like the archaeological record, these traces and mappings are never complete and always highly contingent:

  • Who records? Flickr’s user base has demographic biases.
  • When are photographs taken, in response to which events?
  • Where are photos taken, which locations are popular or which not?
  • What is left behind - what is being photographed?
  • What 'survives'? The geo-referenced photos are only a subset of those originating from a particular city.

To explore some of these questions the timeline interface enables us to select and see different time ranges and 'play' the sequence of locations, visualising the order in which the photos are taken. The rhythms of activity may be as a result of an individual or group social activity, and are mostly played out in public space - the accretion and accumulation of photographic data renders the grid iron of the city street layout visible in the popular centres of cities.


Map Metrics

This release allows us to explore the geography of the data further by selecting different metrics from the 'map' dropdown.

Density
The density view allows us to see more easily where images are clustered on top of each other, which would not otherwise be visible as individual points.
Date Taken
When this metric is selected the points are coloured according to date, allowing us to see which places are photographed at similar times
View Count
This metric is divided into a continuous range of five sections (0 - 4, 5 - 9, 10 - 49, 50 - 99, 100+) These numbers represent the view count up to the end of 2012
Time of Day
We are using the photo timestamp to display when photos are taken and have broken the day into five sections (0 - 5, 6 - 11, 12 - 16, 17 - 20, 21 - 23). This metric is dependent on the accuracy of the timestamp from the camera, so the results are not always accurate - some photos that have a timestamp for a time in the middle of the night have clearly been taken in the day.

Amsterdam and San Francisco Metrics


Tags & Titles

When a Flickr user uploads their photographs, they have the opportunity to give their image a title and apply descriptive keywords or Tags.

These Tags & Titles give us a further way of organising and displaying the photo locations. After all of the data has been loaded for a given search, the top 1000 tags and title words are displayed in the right hand side panel under the Tags tab. The Tags & Titles list can be sorted by Count (Number of photos) or Alphabetically enabling us to more easily navigate the list.

When you click on a tag, every photo that uses that tag is displayed and all of the other photo location points are greyed out. These words have been used to describe the spaces, places, activities and things that Flickr users photograph. Some of the tags are automated by other services like instagram.

San Francisco - Titles Describing Geographical Features

  • San Francisco - 'bridge'
    Bridge
  • San Francisco - 'alcatraz'
    Alcatraz
  • San Francisco - 'beach'
    Beach

London - Tags Describing Geographical and Social Features

  • London - 'thames'
    Thames
  • London - 'canal'
    Canal
  • London - 'stpauls'
    St Pauls
  • London - 'shard'
    Shard
  • London - 'eye'
    Eye
  • London - 'bridge'
    Bridge
  • London - 'march'
    March
  • London - 'occupy'
    Occupy
  • London - 'parade'
    Parade
  • London - 'pride'
    Pride
  • London - 'royalwedding'
    Royal Wedding
  • London - 'riot'
    Riot

Users

We have added a tab which shows the first 1000 Flickr user ID's by numbers of uploaded photos that are geotagged in each city map. This gives us a further way of viewing locations according to who has taken and uploaded a photo. The different spatial configurations of user photo locations very quickly highlight the different usage patterns of how individuals see and respond to the city. The selection of images below are from Berlin in 2012, the metric displayed is time of the day, which gives further insight into when users are active.