Brief Descriptions of Visual Features for Baseline TRECVID Concept Detectors
Akira Yanagawa, Winston H. Hsu, Shih‐Fu Chang, Akira Akira, winston · 2006
We describe the visual features used in the implementations of baseline concept detectors for our participation in NIST TRECVID 2006 high-level feature detection. Such features are part of the rich feature set that has been shown to be effective in previous TRECVID experiments, first in [1] then repeated in [2] and [3]. Original algorithms for extracting each feature may be found in references [4-6]. Publications of research using the feature data included here may acknowledge Akira Yanagawa and Digital Video/Multimedia Lab of Columbia University, and cite this technical report. References to original algorithms, when needed, should be made to those listed in the reference section.