资 源 简 介
The success of Bag-of-visual Word (BoW) scheme has been demonstrated in many research works. Among these works, near-duplicate image/sub-image retrieval is one of the important
applications. SOTU is a toolkit for near-duplicate image and video retrieval as well as detection which is fully based on BoW. SOTU includes routines for
1. Visual vocabulary construction;
2. Vector quantization (VQ) in various manners;
3. Functions for near-duplicate image retrieval and detection.
Our aim to release such a tool is to make people"s works in this area comparable to each other and relieve the efforts of repeating others" work. Techniques integrated with SOTU can be found in our recent works:
[1 ] "On the Annotation of Web Videos by Efficient Near-Duplicate Search", IEEE Trans. on Multimedia, Vol. 12, 2010.
[2 ] "Flip-invariant SIFT for Copy and Object Detection", IEEE Trans. on Image Processing, (to appear).
