Pre-processing feature added

You can now create masks and pre-processed target data-sets for PASCAL based data-sets.

These data-sets are stored in the SuperPixels directory in the root of the PASCAL data-set.

Each pre-processed data-set creates 2 directories: 100_sp_mask and 100_sp_target (assuming you created a 100 superpixel data-set). Each of these sub-directories contains .pt files for loading into your PyTorch model at train/validation/test time.