Module pyVerif.Protocol
Manage protocol files.
The directory is structured as follow:
<TOP>------+-----files_real_ids.lst
|
+-----norm---+----<for_norm_1.lst>
| |
| +----<for_norm_2.lst>
| |
| +----<for_norm_X.lst>
|
+-----<p1>----+----dev-----+----for_models.lst
| | |
| | +----for_models_impostors.lst
| | |
| | +----for_scores.lst
| |
| +----test----+----for_models.lst
| |
| +----for_models_impostors.lst
| |
| +----for_scores.lst
|
+-----<p2>----+----dev-----+----for_models.lst
| | |
| | +----for_models_impostors.lst
| | |
| | +----for_scores.lst
| |
| +----test----+----for_models.lst
| |
| +----for_models_impostors.lst
| |
| +----for_scores.lst
|
+-----<pT>----+----dev-----+----for_models.lst
| |
| +----for_models_impostors.lst
| |
| +----for_scores.lst
|
+----test----+----for_models.lst
|
+----for_models_impostors.lst
|
+----for_scores.lst
where:
- <...> mean some user-specified name,
- <TOP> is the top of the protocol directory,
- files_real_ids.lst is the list of real client id associated to
each file,
- *.lst are lists of files/client associations,
- norm is the directory containing lists of files used for
normalization, like for example the world model,
- <p1>, <p2>, <pT> are some variant of the protocol,
(the variant is optional if you only have one variant),
- dev and test are the mode of use of the protocolk
- for_models.lst are the lists of files we can use for training
models,
- for_models_impostors.lst are the lists of files we can use for
training models, but these are samples of impostors, (this file is
optional),
- for_scores.lst are the lists of files we can use for computing
scores,
Each line of the files *.lst are like (with ids as key):
<filename1> [<filename2> ...] <id1>