Image sequence geolocation with human travel priors

By Evangelos Kalogerakis, Olga Vesselova, James Hays, Alexei A. Efros, Aaron Hertzmann. 2009

174 citations recorded by the source corpus.

This paper presents a method for estimating geographic location for sequences of time-stamped photographs. A prior distribution over travel describes the likelihood of traveling from one location to another during a given time interval. This distribution is based on a training database of 6 million photographs from Flickr.com. An image likelihood for each location is defined by matching a test photograph against the training database. Inferring location for images in a test sequence is then performed using the Forward-Backward algorithm, and the model can be adapted to individual users as well. Using temporal constraints allows our method to geolocate images without recognizable landmarks, and images with no geographic cues whatsoever. This method achieves a substantial performance improvement over the best-available baseline, and geolocates some users' images with near-perfect accuracy.

Subfields: Aerospace Engineering, Computer Vision and Pattern Recognition

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Aerospace Engineering

In Computer Vision and Pattern Recognition

Citations in the corpus

PlaNet - Photo Geolocation with Convolutional Neural Networks (2016, background): “Images2GPS [25] also trains an HMM, but instead of landmarks, its classes are a set of geographical cells partitioning the surface of the earth.”
Clues from the beaten path: Location estimation with bursty sequences of tourist photos (2011, methodology): “The wide availability of online consumer photo collections in recent years has spurred much research in the field [14, 12, 21, 18, 15, 16, 4, 13, 6, 8, 7, 9].”
Worldwide Pose Estimation Using 3D Point Clouds (2012, background): “In follow-up work, human travel priors are used to improve performance for sequences of images [14], but the resulting locations are still coarse.”
Accurate Image Localization Based on Google Maps Street View (2010, methodology): “[7] perform the task by calculating a number of low-level features - such as color histograms and texton histograms - for 6 million images while assuming that there is a very close match for the query image in their dataset.”
Predicting Geo-informative Attributes in Large-Scale Image Collections Using Convolutional Neural Networks (2015, background): “Work in scene classification has considered categories that could be geo-informative, like urban versus rural [40].”
Image Labeling on a Network: Using Social-Network Metadata for Image Classification (2012, methodology): “Another commonly used source of metadata comes directly from the camera, in the form of exif and GPS data [16, 14, 12, 11].”
Learning and Calibrating Per-Location Classifiers for Visual Place Recognition (2013, background): “The output can be a coarse geo-localization on the level of continents and cities (Doersch et al. 2012;Hays andEfros 2008;Kalogerakis et al. 2009) or a name of the depicted landmark (Li et al.”
Retrieving landmark and non-landmark images from community photo collections (2010, background): “[14] build on the previous result by exploiting the time each photo is taken, much like [6].”
Supervised Text-based Geolocation Using Language Models on an Adaptive Grid (2012, background): “For tweet-level geolocation, it will be useful to build on work in geolocation that considers the temporal dimen-sion (Chen and Grauman, 2011; Kalogerakis et al., 2009; Sadilek et al., 2012) to make better predictions for documents/images that are surrounded by others with excellent cues, but which…”
SKYLINE2GPS: Localization in urban canyons using omni-skylines (2010, background): “The use of time-stamped photographs has also proved beneficial for geo-localization [14].”
Travel route recommendation using geotags in photo sharing sites (2010, methodology): “…spatial information into Web applications; recent research efforts have shown the potential of geotags by developing various geotag-associated applications such as georeferenced image search [7, 8], automatic image geolocation [9, 10, 11], and georeferenced content browsing [15, 16, 17, 18, 19].”
Location Recognition Using Prioritized Feature Matching (2010, methodology): “Others have made use of information from image sequences; while this is a common approach in the SLAM (Simultaneous Localization and Mapping) community, human travel priors have also been used to georegister personal photo collections [22].”