How to generate the pickle file on my own dataset?

[CSer-Tang-hao]

Can you share the code or script on how to prepare a given dataset into your pickle-style dataset? Looking forward to your reply, thanks a lot!!!

[MartaYang]

Hi, the key is to generate 3 pair (data-label pair) of files from your dataset. It doesn't matter what format they are in. For example, I simply generate six .npy files ('train_data.npy', 'train_label.npy', 'val_data.npy', 'val_label.npy', 'test_data.npy', 'test_label.npy') for tieredImageNet by appendix-how_to_prepare-tiredimagenet-dataset.ipynb and use these files in tiered_imagenet.py.

[CSer-Tang-hao]

For your CUB-FS file, I want to know whether the data is the original image or the image cropped with provided bounding box?

[MartaYang]

Hi, CUB-FS images are cropped with given bounding boxes following the previous works DeepEMD and FEAT.