Feature extraction is often crucial for problem solving in signal processing and pattern recognition.
The recent research effort in the standardisation and validation of MPEG-7 feature descriptors is just another example. However, it would be interesting to see how MPEG-7 features, especially those originally proposed and tested for image retrieval tasks, scale for semantic analysis of multimedia signal. Our argument is that some schemes of feature combination and classifier combination are necessary.
I would like to share some preliminary results obtained from our experiments on using some visual and audio features for content organisation, scene classification, and music instrument recognition.
Last modified: Thursday, 28-Jul-2005 17:23:30 NZST
This page is maintained by the seminar list administrator.