-------------------------
HOG Features:

There are two archives with the HOG data for the open set object recognition experiments:

HOG_experiments_88.tgz
HOG_experiments_212.tgz

For the article "Towards Open Set Recognition", we did not use fixed partitions - we performed 
random cross validation style training testing over a number of folds. In each directory, there 
is a script to generate random training / testing data in LIBSVM format. For example, to generate 
data for the open world of 212 classes for the positive class "009.bear", you'd run:

$ perl data_set_generator_212.pl -f file-list-testing-212 -c 009.bear -t 1

This will create a directory called 009.bear-1/ with a testing file 'testing_0.027.data', training file 
'training_0.027.data' and a file with just positive features 'training_1-class.data' (for training the
 one-class formulation). The class breakdown will match what is described in the paper. The command syntax 
for the data_set_generator_88.pl script is exactly the same. 

If you want to run several instances of the scripts at the same time, make sure to use unique numerical 
tags via the -t option so you don't overwrite any data.

-------------------------
LBP-like Features for Object Recognition:

There are two archives with the LBP-like features for the object recognition experiments:

lbp-like-feature-vectors-subset-88.tar.gz
lbp-like-feature-vectors-subset-212.tar.gz

The complete set of features for training and testing for all five folds are contained
within the archives.

-------------------------
LBP-like Features for Face Verification:

There is one archive with the LBP-like features for the face verification experiments:

lbp-like-feature-vectors-face.tgz

The complete set of features for training and testing for all five folds are contained
within the archive.

-------------------------
Gabor Features for Face Verification:

This particular data set is extremely large. If you are intersted in these features, please contact
Walter Scheirer (wscheirer@fas.harvard.edu) to arrange a special transfer.

-------------------------