Initialising train_dataset in test.py

[jacobdear1]

Hi, this work is very impressive!

I was running the test.py script and ran into an issue that, when using my own dataset it could not find trainA in my test_directory, but this should be the case as I have split my training and testing data into 2 different folders, and hence when pass 2 different directories test_dir and train_dir respectively to the --dataroot option in python files.

This is my issue, where working_dir needs to be replaced with the path in your system to the UNSB folder;

Traceback (most recent call last):
File "test.py", line 48, in
train_dataset = create_dataset(util.copyconf(opt, phase="train"))
File "/working_dir/UNSB/data/init.py", line 57, in create_dataset
data_loader = CustomDatasetDataLoader(opt)
File "/working_dir/UNSB/data/init.py", line 73, in init
self.dataset = dataset_class(opt)
File "/working_dir/UNSB/data/unaligned_dataset.py", line 35, in init
self.A_paths = sorted(make_dataset(self.dir_A, opt.max_dataset_size)) # load images from '/path/to/data/trainA'
File "/working_dir/UNSB/data/image_folder.py", line 26, in make_dataset
assert os.path.isdir(dir) or os.path.islink(dir), '%s is not a valid directory' % dir
AssertionError: ../test_directory/trainA is not a valid directory

I have removed this line (line 48 in test.py), as shown in the screenshot below and it works as expected. Is there any reason for this, as it seems redundant?