trecvid - varying frame numbering
Yufei raised the following problem:
> We found that the frame numbers obtained from different MPGE decoder are
> slightly different from each other, which will make the boundary of CUT
> So we want to know how the evaluation ground truth will be defined at
> NIST. In addition, if our ground truth is different from NIST's, will
> our results be still valid? Thanks.
We will try to find a way to correctly score cuts inspite of the differences
in frame numbers you have encountered. Arnon Amir has suggested a solution.
But I would like to have more information from Yufei and anyone else who has
seen this phenonmenon:
- what specific files did it happen on?
- what codecs gave different results
- were the differences consistent throughout the files?
- how large were the differences - mean, max
This sort of information will help us understand the size of this problem and
craft a more precise solution.
Paul Over - Retrieval Group
Information Access Division
Information Technology Laboratory
National Institute of Standards and Technology
Bldg. 225 Rm. A211 (Mailstop 8940)
Gaithersburg, MD 20899-8940 USA
Voice: 301 975-6784 Fax: 301 975-5287
Date Index |
Thread Index |
Problems or questions? Contact email@example.com