Given a query image of an object, our objective is to retrieve all instances of that object in a large (1M+) image database. We adopt the bag-of-visual-words architecture which ha...
Ondrej Chum, James Philbin, Josef Sivic, Michael I...
Abstract. Humans can associate vision and language modalities and thus generate mental imagery, i.e. visual images, from linguistic input in an environment of unlimited inflowing i...
Near-duplicate image detection is a critical task in copyright protection. More challenging than the common similarity search, this task requires not only the retrieval of the top...
In this paper, we present a fast algorithm for filling unknown regions in an image using the strategy of exemplarmatching. Unlike the original exemplar-based method using exhaustiv...