Images no longer available

[cmishra]

Hi,

I've tried to use this toolbox to download the full set of images. I've gotten ~4.5 million images, but a significant portion are still left (distributed roughly evenly over all 100 data subsets).

Upon manual inspection, many of these images are no longer being hosted while a minority have bot-shielding strategies up which a simple header addition didn't defeat. More seem to fall into the former category.

Are there any alternate sources for missing images? A torrent or tarball download? If not, can we at least temporarily set one up? It would be relatively inexpensive.

If I can get a copy of the dataset, I may be able to convince certain folks to host it long term.

[sbrodehl]

Any news here?

[cmishra]

Nothing ever came of this. I emailed the paper authors directly as well.

[GabrielLin]

@cmishra Do the authors reply to you? Thank.

[cmishra]

Nope.

…

On Wed, Dec 4, 2019, 2:24 AM GabrielLin ***@***.***> wrote: @cmishra <https://github.com/cmishra> Do the authors reply to you? Thank. — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#3?email_source=notifications&email_token=AA4J5KJPCJ75YZBDRLWUKG3QW5LK7A5CNFSM4C3INLEKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEF4AEEY#issuecomment-561512979>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AA4J5KPTS6JJVPMYD3THOFTQW5LK7ANCNFSM4C3INLEA> .