Hi Everyone !,
I just joined the forums, though I have been using fastai for a couple of weeks.
I had a couple of doubts and would greatly appreciate if anyone could give me clarity on it.
a)Similar to the get_transforms() function
data = ImageDataBunch.from_csv(path, ds_tfms=tfms, size=24);
that you can apply on the Training Dataset, can I also do the same for Test Dataset, ( If Iam adding the Test dataset just before prediction ? )
learn.data.add_test(ImageList.from_folder(test_folder))
Does not see m to have the option to add tfms() in the parameters to the add_test() method. Or does the tfms get applied to the test dataset by default too ?
eg: If i do a p_lighting or contrast on the train dataset, I assume I should also do the same on the test dataset before scoring the test images ?
b) If I load the test dataset from a test folder , and predict using the .get_preds method,
preds,y = learn.get_preds(DatasetType.Test)
how do I know which Test image file corresponding to the values that are stored in the ‘preds’ list ? What is the order of test images? ( is it sorted alphabetically before sending it in batches ? )
Thanks