Photomapping: localization with images
Institution: | California State University – Northridge |
---|---|
Department: | Department of Computer Science |
Degree: | MS |
Year: | 2015 |
Keywords: | Image localization; Dissertations, Academic – CSUN – Computer Science. |
Record ID: | 2060793 |
Full text PDF: | http://hdl.handle.net/10211.3/134467 |
With the prevalence of smart phones in our daily lives, sharing images has become an important part of social networking and communication. Along with sharing a photo, users often include a tag, short text related to the subject of the image. When someone uploads a photo to an image sharing website, with a tag indicating the location where the photo was taken, are these user submitted location tags accurate? This thesis examines the accuracy of geotagging of social media content, specifically of photos. Geo-location metadata from photos along with reverse geocoding of latitude and longitude coordinates is used to analyze the geotagging phenomena and its accuracy. It then proposes an approach for auto geotagging images, leveraging computer vision???s object recognition techniques. Specifically, images with the presence of a city landmark are used to assign a geotag to the photos. The accuracy of the auto geotagging is determined using a large of set of photos taken from Flickr.