|
MMM is a top-ranking annual forum for researchers working in
the multimedia area, and the MMM 08 was held in Kyoto,
Japan, from 11 to 14 January 2008.
|
Neo Shi Yong and
Zheng Yan-Tao |
The award-winning paper was the work of two graduate
students from NUS School of Computing, Zheng Yan-Tao and Neo
Shi-Yong. Their co-authors were Professor Chua Tat-Seng from
the School, and Dr Tian Qi from the Institute of Infocomm
Research (IČR) in Singapore.
The paper addresses the problem of unsatisfactory
performance of object-based image retrieval systems.
Identifying the cause of the problem as the systems’
over-dependence on visual similarity and regularity among
images in the same semantic class, the paper proposes a
novel image presentation, i.e., bag of visual synset.
The paper presents the results of testing on Caltech-256
datasets, showing that by fusing visual words in a relevance
consistent way, the visual synset can partially bridge
visual differences of images of same the class and deliver
satisfactory retrieval of relevant images with different
visual appearances.
The paper is published in Lectures Notes on Computer
Science, Volume 4903 (2008). Find out more about it
here. |
