An extract from our databset of 30,000 images covering 20km of terrain

In this project we looked at the problem of location recognition in a large image dataset. This involved finding the location of a query image in a large geotagged dataset containing 30,000 streetside images of a city. We investigate the performance of the vocabulary tree approach as the size of the database grows. In particular, we show that by carefully selecting the vocabulary using only the most informative features, retrieval performance is significantly improved. We also present a generalisation of the standard vocabulary tree search algorithm which improves performance by considering several likely paths.

Our tree search algorithm considers the N best nodes at each level (left to right N = 1, 2, 5, 9). Cells are coloured from red to green according to the depth at which they are searched, while gray cells are never searched.