CERTH/CEA LIST at MediaEval Placing Task 2015

We describe the participation of the CERTH/CEA LIST team in the Placing Task of MediaEval 2015. We submitted five runs in total to the Locale-based placing sub-task, providing the estimated locations for the test set released by the organisers. Out of five runs, two are based solely on textual information, using feature selection and weighting methods over an existing language model-based approach. One is based on visual content, using geo-spatial clustering over the most visually similar images, and two runs are based on hybrid approaches, using both visual and textual cues from the images. The best results (median error 22km, 27.5% at 1km) were obtained when both visual and textual features are combined, using external data for training.