Need some help with VIPER-PE

[Christian M Netter]

I have read the documentation, but I'm still not sure how exactly to
use VIPER-PE. Is there a complete example that I can download? I.e.
including ground truth, candidate data, parameter file, etc. etc. +
exact command line for execution. I think this would be a great help
to me...

Thanks.

Chris.

[David Mihalcik]

I started writing a case study document, but it doesn't link to any
sample files to download. I'll update the page to use the new file
format and link to some samples.

http://viper-toolkit.sourceforge.net/docs/cases/

[David Mihalcik]

I've put some example files up at:

http://viper-toolkit.sourceforge.net/stuff/text-word/

To run this, use the command:

java -jar viper-pe.jar -epf object-all.epf -g sample-lamp.xgtf -r
lamp-text-word.xgtf -pr textdetect.pr

Basically, this does a simple evaluation of overlay text in a short
video (the lamp-moving sample available on the viper home page) where
the target/truth is marked at the line level, while the
candidate/result data is marked at the word level. This evaluation
appropriately gives a one-many match, instead of using a hungarian
match, which would only match the largest word to the line (in this
case, 'laboratory').

If you want to do a framewise evaluation, which probably makes more
sense for text detection than the object evaluation, you can substitute
this for the evaluation block:

#BEGIN_FRAMEWISE_EVALUATION

OBJECT Text [- -]
Location : matchedpixels missedpixels falsepixels \
fragmentation arearecall areaprecision \
[arearecall .7] [areaprecision .7]

#END_FRAMEWISE_EVALUATION