Loader bug
xialeiliu opened this issue · 7 comments
Could you confirm that in the following line, self.shot_num should be self.way_num?It is not consistent with mini-imagenet-tools and gives error.
https://github.com/yaoyao-liu/tiered-imagenet-tools/blob/master/tiered_imagenet_dataloader.py#L189
Thanks for reporting this issue. I have updated the code.
Thanks for your quick reply.
Another question about using import scipy.misc as scm,
in mini-imagenet-tools, import imageio is used, I am wondering what's the reason to change it?
For me it seems imageio works fine and scm gives me problems.
In the initial versions for the two projects, I use similar libraries.
After that, I updated the mini-imagenet-tools more often, but I didn't update this project correspondingly. Sorry about that.
You may directly use the dataloader for miniImageNet on tieredImageNet with minor changes. You may also apply the dataloader I used in this project for different datasets.
Thanks, good to know your new project.
And last thing is that the following line works for me using:
class_idxs = list(range(0, self.way_num))
https://github.com/yaoyao-liu/tiered-imagenet-tools/blob/master/tiered_imagenet_dataloader.py#L147
with these fixes, it works fine for me now.
Thanks.
You may create a pull request for this issue so that I can include you in the contributors' list.
It's fine, only several minor points, you could fix it directly if that is correct to you.
Thanks.
Thanks for the contributions.