Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Missing data in birdsnap/test #17

Open
KarolinaPondel opened this issue Aug 7, 2024 · 1 comment
Open

Missing data in birdsnap/test #17

KarolinaPondel opened this issue Aug 7, 2024 · 1 comment

Comments

@KarolinaPondel
Copy link

I tried to run src/notebooks/zeroshot.ipynb, but but unfortunately after loading the data (dataloader) there is a
'''
FileNotFoundError: Found no valid file for the classes 046, 066, 123, 299, 302, 351, 403, 436, 465. Supported extensions are: .jpg, .jpeg, .png, .ppm, .bmp, .pgm, .tif, .tiff, .webp.
'''
However, it is possible to run the rest of the code despite this and I get a pretty good result, strange.... In the second scenario I removed the missing folders, everything went through without error, however in the results the bird images shown have mixed up captions....

@fuersta
Copy link
Member

fuersta commented Aug 8, 2024

Yes, unfortunately there are some classes of birdsnap missing, which are not available to download anymore.
The torchvision package we used during development did allow empty classes, with the current version you have to set a flag in the dataset called allow_empty.
https://pytorch.org/vision/main/generated/torchvision.datasets.ImageFolder.html
If you remove the empty folders, you get wrong class allocation leading to wrong results!

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants