Skip to content

Conversation

rvandeghen
Copy link
Contributor

Solve the problem of (#5913)
@datumbox

Solve the problem of (pytorch#5913)
@datumbox
Copy link
Contributor

@rvandeghen I don't believe this needs fixing. The original issue with PIL and floats has been already addressed on the latest main branch with #5596.

I suspect you are facing issues because you are using reference scripts from main (v0.13.0-dev) using v0.12.0 which doesn't contain the above patch. Could you try the latest nightly and confirm?

@rvandeghen
Copy link
Contributor Author

@datumbox Yes my bad.
I will close the PR.
However, I am always building on the reference script for detection for my own research so I always try to improve the whole pipeline.
Is it something that could interest you ?

@datumbox
Copy link
Contributor

@rvandeghen This specific issue was addressed with the other PR by making pad for PIL work with float values. So converting the scripts to use ints is no longer needed on the latest version. Nevertheless, we are happy to receive other improvements if you spot further issues. Feel free to open a new issue and let's chat there. :)

@rvandeghen rvandeghen closed this May 10, 2022
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

3 participants