SVHN svhn_dsprepare error

[pr]

Hello,

I've been following SVHN tutorial (http://eblearn.sourceforge.net/svhn_tutorial.html#prepare_the_dataset) along and I've had an error at point 3 in Prepare the dataset:

Error: cannot select an empty idx idx that is a scalar (<scalar>), in ebl::idxspec::select_into at ..\..\libidx\src\idxspec.cpp:459

and dssplit.exe stopped working.

Do you have any idea what might be wrong?

Thanks in advance.

___________________________________________________________________

             Dataset splitter for libeblearn library

___________________________________________________________________

input parameters:

  input directory: svhn

  input dataset name: svhn_ynuv7_extra

  output dataset 1 name: svhn_ynuv7_extra_val

  output dataset 2 name: svhn_ynuv7_extra_train

  max per class for dataset 1: 200 (the rest goes to dataset 2)

  number of dataset pairs to randomly draw: 1

___________________________________________________________________

Data matrix type: long

Setting dataset name to: svhn_ynuv7_extra

Warning: pid() not implemented for Windows

.

Loading dataset svhn_ynuv7_extra from svhn/svhn_ynuv7_extra_*.mat

Loaded svhn/svhn_ynuv7_extra_data.mat ([531131, from <empty> to <empty>])

Loaded svhn/svhn_ynuv7_extra_labels.mat (531131)

Loaded svhn//svhn_ynuv7_extra_scales.mat (531131)

warning: failed to load dataset file svhn/svhn_ynuv7_extra_jitters.mat

Loaded svhn/svhn_ynuv7_extra_classes.mat (10x2)

Dataset "svhn_ynuv7_extra" contains 531131 samples (of dimensions <empty> and long precision), distributed in 10 classes

: 45550 "0", 90560 "1", 74740 "2", 60765 "3", 50633 "4", 53490 "5", 41582 "6", 43997 "7", 35358 "8", 34456 9".

Setting dataset name to: svhn_ynuv7_extra_val

Warning: pid() not implemented for Windows

.

Setting dataset name to: svhn_ynuv7_extra_train

Warning: pid() not implemented for Windows

.

Setting target dimensions to <empty>

Setting target dimensions to <empty>

Splitting "svhn_ynuv7_extra" into datasets "svhn_ynuv7_extra_val" and "svhn_ynuv7_extra_train", limiting dataset "svhn_y

nuv7_extra_val" to 200 samples per class, the rest going to "svhn_ynuv7_extra_train".

Max number of samples per class: 200

Input data samples: [531131, from <empty> to <empty>]

Dataset "svhn_ynuv7_extra_val" will have 531131 samples of size <empty>

data matrix is [empty midx]

Dataset "svhn_ynuv7_extra_train" will have 531131 samples of size <empty>

data matrix is [empty midx]

Adding data to "svhn_ynuv7_extra_val" and "svhn_ynuv7_extra_train".

(original index 12500) 1 / 2000: add sample as 0 (0) eta: 0s elapsed: 0s (dataset 1)

Error: cannot select an empty idx idx that is a scalar (<scalar>), in ebl::idxspec::select_into at ..\..\libidx\src\idxs

pec.cpp:459