turn from: http://blog.csdn.net/liuheng0111/article/details/72772922
Inria Holidays
Inria Holidays Dataset,herve Jegou, a dataset that is used by their institute on a regular vacation, is a total of 1491 images, 500 query (one group) and 991 related images, and 128-D Sift points have been extracted from 4,455,091, visual dictionaries from flickr60k, links.
Oxford Buildings
The Vgg group of the Oxford buildings Dataset,oxford collection of 5062 building images from Flickr, links.
Oxford Paris
The Vgg group of the Paris Dataset,oxford collected 6412 photos of Paris tours from Flickr, including Eiffel Tower and links.
201Books and CTurin180
The CTurin180 and 201Books data sets,2011.5,telecom Italia is provided in the compact descriptors for Visual Search, which includes: Nokia E7 shot 201 The cover of the book (from multiple angles, each 6), a total of 1.3GB Turin City 180 buildings of video images, shooting camera has Galaxy S, IPhone 3, Canon A410, Canon S5 is, a total of 2.7GB, links.
Stanford Mobile Visual Search
Stanford Mobile Visual Search Dataset,2011.2,stanford, including 8 scenarios, such as CD cover, painting, and so on, each group of related pictures are collected from different cameras (mobile phone), all scenes a total of 500 pictures, links A patch dataset was released later, and the Compact descriptors for the Visual Search patches Dataset, proofreading the same patch.
Ukbench
Ukbench Database,2006.7,henrik Stewénius The data set provided in his CVPR06 article, the images are 640*480, each group has 4 charts, the file is close to 2GB, providing visual words, links.
Mir-flickr
Mir-flickr-1m,2010,1m photos on FLICKR also provide a 25K subset of downloads, links.