Invited Talk by Prof. Jiri Matas: Image Retrieval in Large Collections - Beyond Similarity Search
- Speaker: Prof. Jiri Matas
- Date: Thursday, January 08, 2015
- Time: 14:15-15:15
- Location: Seminar room MI 03.13.010, TUM Informatics Faculty, Garching
Traditionally, image retrieval is formulated as the search for the most
similar image or sub-image in the database. In large collections, such searches
often produce results of limited interest, returning images with no new
information and virtually identical content.
We propose retrieving the most dissimilar
images of the object depicted in
the query and develop two novel methods for a particular instances of
"dissimilarity search": the zoom-in and zoom-in for finding the maximum and
minimum spatial resolution depicting the query. Intuitively, the searches
answer the "What is this?" and "Where is this?" questions. Furthermore, two
problems straddling the boundary between image retrieval and data mining are
formulated: for every pixel in the query image, (i) find the database image
with the maximum resolution depicting the pixel and (ii) find the frequency
with which it is photographed in detail.
An efficient and reliable solution is proposed based on two novel techniques,
the hierarchical query expansion that exploits the document at a time (DAAT)
inverted file and a geometric consistency verification sufficiently robust to
prevent topic drift within a zooming search.
Experiments show that the proposed method finds surprisingly fine details on
landmarks, even those that are hardly noticeable for humans.
Jiri Matas is a full professor at the Center for Machine Perception, Czech
Technical University in Prague. He holds a PhD?
degree from the University of
Surrey, UK (1995). He has published more than 200 papers in refereed journals
and conferences. Google Scholar reports about 20 000 citations to his work and
and an h-index above 50.
He received the best paper prize at the British Machine Vision Conferences in
2002 and 2005, at the Asian Conference on Computer Vision in 2007, at the
Scandinavian Conference on Image Analysis and at the Image and Vision Computing
New Zealand Conference in 2013. His students received a number of awards, e.g.
Best Student paper at ICDAR 2013, Google Fellowship 2013, and various "Best
J. Matas is on the editorial board of IJCV and was the Associate Editor-in-Chief
of IEEE T. PAMI. He is a member of the ERC Computer Science and Informatics
panel. He has served in various roles at major international conferences, e.g.
ICCV, CVPR, ICPR, NIPS, ECCV, co-chairing ECCV 2004 and CVPR 2007. He is a
program co-chair for ECCV 2016.
His research interests include object recognition, image retrieval, tracking,
sequential pattern recognition, invariant feature detection, and Hough Transform
and RANSAC-type optimization.
For more, see http://cmp.felk.cvut.cz/~matas