Pretrained Model Output Permuted about places365 HOT 2 CLOSED

hmmm, is the result consistent with the testing script (https://github.com/CSAILVision/places365/blob/master/test_placesCNN_pytorch.py)? The output of the testing script seems all-right.

from places365.

It appears the ImageFolder dataset creator is sorting the classes in a way I did not expect.

classes = sorted([str(i) for i in range(365)])
yields the ordering
['0', '1', '10', '100', '101', '102', '103', '104', '105', '106', '107', '108', '109', '11', '110', '111', '112' ,...]
which is the behavior I observe.

Thus this is caused by the ImageFolder dataset creator.

from places365.