-
Notifications
You must be signed in to change notification settings - Fork 214
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
Changes for 0.9.0 release #268
Merged
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
`f"{req.project_name}{req.specifier}"` is more generic than `f"{req.project_name}{req.specs[0][0]{req.specs[0][1]}"`. Either solution fix the issue.
#253 Dependency record is not required by Bioimage.IO. Use `generate_default_deps=False` in `export_bioimageio()` by default to streamline the export process and enhance clarity for new users.
In some cases need to add extra context to prevent overlapping write regions of non-neighboring blocks. Cf. https://forum.image.sc/t/trouble-using-stardist-predict-instances-big/88871/6
windows-latest with tensorflow 2.15.0 yields different results ¯\_(ツ)_/¯
Negative values in label masks will turn off all losses for that pixel during training. This is especially useful for training with sparse annotations, i.e. not all pixels of an image have to be annotated. This is similar to https://github.com/ksugar/stardist-sparse, but all negative label pixels are ignored (not just -1), and the object classification loss is also disabled (if 'n_classes' is not None).
To disregard pixels that have been intentionally set to -1
Fix and improve the export function for Bioimage.IO
- Requires Keras 3-ready csbdeep package (to be released) - Use abstracted BACKEND to replace keras.backend Work in progress...
Seems like Keras 3 is more strict about some mixed data type arithmetic operations -> cast ground truth and intermediate variables to float data type. Not all of these casts are likely necessary, but they shouldn't hurt either.
The parameters "workers" and "use_multiprocessing" have moved from `model.fit` to the data generator (based on `keras.utils.PyDataset`). Also, the data generator apparently must return tuples (and not lists)? -> https://stackoverflow.com/a/78158487
Model export seemingly not possible with Keras 3.
Python 3.12 apparently removed distutils
No longer use environment variable SETUPTOOLS_USE_DISTUTILS, which doesn't work with Python 3.12. Don't remember why we had to add this in the first place...
Return lists in data generator
numpy.distutils has been removed in Python 3.12 https://numpy.org/doc/stable/reference/distutils_status_migration.html#distutils-status-migration
Add support for Keras 3
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
No description provided.