Skip to content

Exclude the Platynereis neuropil from the loss with an ignore label - #1365

Merged
anwai98 merged 2 commits into
devfrom
platynereis-neuropil-ignore-label
Sep 6, 2026
Merged

anwai98 merged 2 commits into
devfrom
platynereis-neuropil-ignore-label

Conversation

@anwai98

@anwai98 anwai98 commented Sep 4, 2026

Copy link
Copy Markdown
Collaborator

Tagged with constantinpape/torch-em#796. What do you think @constantinpape? Could you please give it a quick review?

@anwai98

anwai98 commented Sep 5, 2026

Copy link
Copy Markdown
Collaborator Author

The CIs are broken because we'd need a new torch-em with the new changes. Let's not worry about those details for now!

@anwai98

anwai98 commented Sep 5, 2026

Copy link
Copy Markdown
Collaborator Author

Any thoughts on this? @constantinpape

@constantinpape constantinpape left a comment

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Idea is good, but I find 255 very dangerous as ignore id.

Comment thread micro_sam/v2/transforms/labels.py Outdated
Comment thread micro_sam/v2/transforms/labels.py
@anwai98

anwai98 commented Sep 6, 2026

Copy link
Copy Markdown
Collaborator Author

Okay this is gonna stay open because I need a few more moving parts figured out in torch-em. Then I'll make a new release there, merge this, and then we are good to fly. Thanks @constantinpape!

@anwai98

anwai98 commented Sep 6, 2026

Copy link
Copy Markdown
Collaborator Author

Gonna merge this. The CIs would break, but I don't wanna wait for torch-em release (because I'm still cross checking all datasets one final time)

@anwai98
anwai98 merged commit 11ab239 into dev Sep 6, 2026
@anwai98
anwai98 deleted the platynereis-neuropil-ignore-label branch September 6, 2026 17:17
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

Successfully merging this pull request may close these issues.

2 participants