diffusiondb icon indicating copy to clipboard operation
diffusiondb copied to clipboard

"download -z" unzips all the images to the same directory

Open shri-zz opened this issue 2 years ago • 0 comments

I ran python download.py -i 1 -r 5 -z based on https://huggingface.co/datasets/poloclub/diffusiondb#downloading-to-a-specific-directory. It downloaded the five zip files as images/part-00000<N>.zip. However, the unzipped images were all in the current directory. Shouldn't they be created in five separate subdirectories? Otherwise, you end up with a single directory with 2M files. https://huggingface.co/datasets/poloclub/diffusiondb says "The 2 million images in DiffusionDB 2M are split into 2,000 folders", and dowload.py is not implementing that intent.

shri-zz avatar Oct 16 '23 04:10 shri-zz