The various tasks described above—content-based image retrieval, object-level image subregion querying, object localization, cut detection, and semantic organization of an image collection — become especially challenging when the image database i