WIP: Simplify and replace postprocessing #199
Open
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
|
Moving off until the next release as we need a rebuild with the Tornado 5 fixes, the Distributed Array Store, and newer dependency versions. Also this code needs some exploration on data to understand what works well, which will be easier to do with a new release. Adding the |
jakirkham
force-pushed
the
simp_n_repl_postprocessing
branch
from
91cf82d to
ce9ee14
Compare
Make sure to import `dask_ndmeasure` along with the other dask-image org libraries. Will be needed to construct the label image. IOW will be used for ROI extraction.
Simply use a threshold with no other constraints.
Also make the chunks concrete (as they are undefined) to allow storage.
jakirkham
force-pushed
the
simp_n_repl_postprocessing
branch
from
735f45c to
8ce5237
Compare
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Outdated suggestions cannot be applied.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.
Rewrites the postprocessing step to use a Dask-based algorithm that is a significant departure from the original postprocessing algorithm. This is done for a few reasons. First the old postprocessing algorithm was pretty sensitive and had many failure modes. Second the previous algorithm had a lot of knobs, which slowed down finding a good solution. Third a lot of heuristics were baked into the old postprocessing algorithm, which made it hard to rationalize or understand what it was doing let alone be confident in the result it found. Borrows some pieces from PR ( #110 ) to do this.