Does not process images

[adbugger]

Running the following command:
python tiered_imagenet_generator.py --tar_dir [your_path_of_the_ILSVRC2012_img_train.tar]

creates the directory structure: tiered-imagenet/train/n**/ but no files.
Further, the processed_images directory is not created.

I tried the same command with the mini-imagenet-tools repo and it works correctly, so I'm assuming that my tar-file is fine.

[yaoyao-liu]

I am very sorry that I don't have time to check this repo. You may directly download processed image using following link:
https://meta-transfer-learning.yyliu.net/download/

[akinaka0x]

I had the same problem.
Replacing line 82 with image_path = glob(os.path.join(img_dir, split, keys, '*')) worked for me.

[yaoyao-liu]

I had the same problem.
Replacing line 82 with image_path = glob(os.path.join(img_dir, split, keys, '*')) worked for me.

Thanks for your contribution. You may create a pull request and I'll merge it.

[akinaka0x]

Replacing line 82 with image_path = glob(os.path.join(img_dir, split, keys, '*')) worked for me.

It turned out the code I posted above was wrong, and I am really sorry for the confusion.
The code I posted above worked for train dataset by accident and does not work for val and test sets of the tiered-ImageNet dataset.

The original code is correct, and the reason I had the problem was that I wrongly specified the --imagenet_dir.
It seems the ImageNet directory structure is assumed to be like this in the program:

[--imagenet_dir] --- n01440764 --- *.JPEG
                  |             |- *.JPEG
                  |
                  |
                  |- n01443537 --- *.JPEG
                  |             |-

However, my ImageNet directory structure was like ILSVRC2012/CLS-LOC/train/n01440764/*.JPEG (I did not use the tar file) and I wrongly specified the --imagenet_dir as ILSVRC2012/CLS-LOC.
In my case, specifying --imagenet_dir as ILSVRC2012/CLS-LOC/train is correct and worked fine.

[yaoyao-liu]

Replacing line 82 with image_path = glob(os.path.join(img_dir, split, keys, '*')) worked for me.

It turned out the code I posted above was wrong, and I am really sorry for the confusion.
In my case, the posted code worked for train dataset by accident and does not work for val and test sets of the tiered-ImageNet dataset.

The original code is correct, and the reason I had the problem was that I wrongly specified the --imagenet_dir.
It seems the ImageNet directory structure is assumed to be like this in the program:

[--imagenet_dir] --- n01440764 --- *.JPEG
                  |             |- *.JPEG
                  |
                  |
                  |- n01443537 --- *.JPEG
                  |             |-

However, my ImageNet directory structure was like ILSVRC2012/CLS-LOC/train/n01440764/*.JPEG (I did not use the tar file) and I wrongly specified the --imagenet_dir as ILSVRC2012/CLS-LOC.
In my case, specifying --imagenet_dir as ILSVRC2012/CLS-LOC/train is correct and worked fine.

Thanks all the same for your interest in our work. I'll check this repo later when I have time.

[akinaka0x]

This code is very useful, and thank you for sharing it with us!