Communications of the ACM
Flickr Photo Data sed to Predict People's Locations
A new algorithm can predict a person's location based on the photos he or she uploads to Flickr.
Credit: Petr Kratochvil
Researchers from University College in England have developed an algorithm that can predict people's location based on the photos they upload to the Flickr file-sharing website.
Using photos shared by 16,000 people in the U.K., the team created a database of 8 million images. The algorithm accessed the photos' global-positioning system and time-stamp data to note all the locations where pictures had been taken by a single camera, as well as to predict where people would take photos in the future based on their past movements.
The team tested the algorithm by comparing their results with a government survey taken to better understand national travel patterns. The researchers report their results and the survey agreed 92 percent of the time. They say they also could focus on individuals, or at least their cameras, and anticipate where that person might be at any given moment.
The research could provide the government with a new way to track people's movement, which could be beneficial for road-building plans and other transportation projects. However, it also raises privacy issues about shared photographs.
From Phys.org
View Full Article
Abstracts Copyright © 2015 Information Inc., Bethesda, Maryland, USA
No entries found