Skip to content

Conversation

@liopeer
Copy link
Contributor

@liopeer liopeer commented Sep 16, 2025

What has changed and why?

  • added ScaleJitter transform
  • made transforms step/epoch aware so that we can have a stop_step/stop_epoch
  • also made some small changes to semantic segmentation transform. we can also leave these away for now and refactor later (the refactoring is anyway not complete with this PR, so might make sense to do everything separately)

What remains:

  • RandomIoUCrop

How has it been tested?

  • unit tests (might make sense to add more sophisticated ones before merging)

Did you update CHANGELOG.md?

  • Yes
  • Not needed (internal change)

Did you update the documentation?

  • Yes
  • Not needed (internal change without effects for user)

liopeer added 30 commits August 7, 2025 21:51
… into lionel-trn-758-add-object-detection-dataset
… into lionel-trn-758-add-object-detection-dataset
… into lionel-trn-758-add-object-detection-dataset
…com:lightly-ai/lightly-train into lionel-trn-758-add-object-detection-dataset
Copy link
Contributor

@stegmuel stegmuel left a comment

Choose a reason for hiding this comment

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

First review.

Copy link
Contributor

@stegmuel stegmuel left a comment

Choose a reason for hiding this comment

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

Left some questions/remarks.

…t2' of github.com:lightly-ai/lightly-train into lionel-trn-1520-implement-object-detection-transforms-pt2
Copy link
Contributor

@stegmuel stegmuel left a comment

Choose a reason for hiding this comment

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

LGTM

@liopeer liopeer merged commit aae1940 into main Sep 29, 2025
16 checks passed
@liopeer liopeer deleted the lionel-trn-1520-implement-object-detection-transforms-pt2 branch September 29, 2025 16:38
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.

4 participants