IIT-KGP Visual Saliency
Data
This page contains a link to the groundtruth data
prepared from Visual Saliency Experimentation, conducted during 2009 January to
March. This data have been used as benchmark data for comparing different
algorithms to determine salient points in an image. Following the links from
these pages, one may obtain the set of original images and their corresponding initial groundtruths and final groundtruths
(as bilevel images) in three different zipped files.
Correspondences are established by tagging the same number in their file names.
Salient locations are recorded for 100 images assisted by 62 volunteers.
Each of these volunteers is shown 24 images.
Images for our experiment are chosen from a larger set of databases
taken from iLab image database1,2, UCID3, Zurich natural
image database4, and the Internet.
1 L. Itti, C. Koch, and E. Niebur, A model of
saliency-based visual attention for rapid scene analysis, IEEE Transactions
on Pattern Analysis and Machine Intelligence, vol. 20, no. 11, pages 1254-1259,
1998.
2 L. Itti, and C. Koch, Feature combination
strategies for saliency-based visual attention systems, Journal of
Electronic Imaging, vol. 10, issue 1, pages 161-169, 2001.
3
G. Schaefer, and M. Stitch, UCID - an uncompressed color image
database, Proc. of SPIE Storage and Retrieval Methods and Applications for
Multimedia, vol. 5307, pages 472-480, 2004.
4 H. P.
Frey, P. Konig, and W. Einhauser, The role of first- and second- order
stimulus features for human overt attention, Perception and Psychophysics,
vol. 69, pages 153-161, 2007.