Closed GaoJiqiang closed 3 years ago
Hi @GaoJiqiang, you have to first download ImageNet before running the code (the code won't download it automatically for you).
Basically you have to obtain a copy of ImageNet (http://www.image-net.org/index), then preprocess the val directory to look like the train directory by running this script. Finally, point --data
to the directory containing the train
and val
folders of ImageNet that you downloaded and processed.
Hope this helps!
I want to train a standard model on imagenet, but there has a problem when downloading the dataset.