facebookresearch / deit

Official DeiT repository
Apache License 2.0
4.02k stars 552 forks source link

ImageNet21K data preparation for pre-training #219

Open mxjecho opened 1 year ago

mxjecho commented 1 year ago

Hi, I want to try imagenet21k pretraining. Could you please tell me how to process the raw data of imagenet21k?

TouvronHugo commented 1 year ago

Hi @mxjecho, Thanks for your message. You mean which data augmentation used? Best, Hugo

mxjecho commented 1 year ago

No, I want to know if I need to do pre-processing on the dataset. I use the winter21 imagenet 21k dataset.

TouvronHugo commented 1 year ago

We didn't do any particular preprocessing, we re-use the version we have internally of the dataset which should corresponds to the official version of the ImageNet dataset

mxjecho commented 1 year ago

Get it. BTW, what is the 'num_class' of Imagenet 21k?

TouvronHugo commented 1 year ago

21844 if I'm not wrong