Skip to content

VQA v2.0 40000+ images are missing #18

Open

Description

@Omair-S

The number of images on the homepage (https://visualqa.org/download.html ) does not match the ones I'm getting after extracting the files. I'm getting this issue only for balanced real images, not for abstract scenes.

Training Images: Expected 82,783 vs. Actual 41,635
Validation Images: Expected 40,504 vs. Actual 40,493 (close match)
Testing Images: Expected 81,434 vs. Actual 41,915

I didn't receive any error while installing or extracting the zip files, so I don't think there is an issue with file extraction. Is it possible that there is some corrupted data in the file which I might not be aware of? Or maybe the file explorer isn't giving an accurate number (is this possible?). I want to confirm if anyone has faced such an issue before with this dataset, or in general while downloading a huge dataset of images locally.

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions