[
Lists Home |
Date Index |
Thread Index
]
Elliotte Rusty Harold wrote:
> At 11:02 AM +0100 2/27/03, Robin Berjon wrote:
>> I'm all for finding a sweet spot but pulling random numbers out of a
>> hat and making broad assumptions about size vs computation won't
>> contribute much in getting there. I am talking about empirically
>> proven, tested, retested, put to work in a wide variety of situations,
>> factors of 10, 20 or 50 (or more, but testing on SOAP is cheating ;).
>
> Is there any actual publication with verifiable, reproducible data? I
> don't need fancy peer-reviewed journals or anything, just a web page
> that outlines the test methodology and provides the code used and the
> test cases.
The ISO/MPEG membership quite naturally tested the technology for its own needs,
though I'm afraid it might not be available.
You can test it yourself partly from the software on the CD that ships with the
Wiley book "Introduction to MPEG-7" but it is quite antiquated and won't have
the kind of performance that is available today. It covers solely BiM and not
further progress beyond it.
Up to now I've mostly been concerned with testing it in-house and to customers
but I'm preparing publicly available and reproducible data. The numbers I quote
are based on the next release of our SDK which is to ship next month. As soon as
it comes out we will make a demo available and I will provide reproducible tests
with everything needed to validate them. Feel free to suggest specific tests you
would like to see there. I'd be happy to announce it here if there is interest.
In the meantime I can certainly dig up benchmarks, but you won't be able to
reproduce them.
> So far, it all sounds like numbers out of a hat to me.
I don't see what my interest would be in that :)
--
Robin Berjon <robin.berjon@expway.fr>
Research Engineer, Expway http://expway.fr/
7FC0 6F5F D864 EFB8 08CE 8E74 58E6 D5DB 4889 2488
|
