Skip to content
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

Segment counts #360

Draft
wants to merge 61 commits into
base: master
Choose a base branch
from
Draft

Segment counts #360

wants to merge 61 commits into from

Conversation

hanslovsky
Copy link
Collaborator

@hanslovsky hanslovsky commented Dec 12, 2019

This is a WIP draft PR to track work on segment counts and how to expose them through the UI to navigate to the (next) smallest segment.

This will only be merged after #343

Note that this does not work for label Paintera datasets (yet)
The actual type of the FragmentSegmentAssignment does not need to be specified. This makes relative datasets a lot easier!
Add remarks to add UI configuration for offset/resolution in the future
The number of channels is not the same as the number of mipmap levels!
The backend should not hold on to the source. Instead, it should create it on demand.

WIP because of weird compilation error.
… dataset

To make datasets really relative to project directory, label-block lookup has to be relative to containing Paintera dataset.
Even on a small CREMI dataset, extracting count features is pretty slow already. Use caches instead.
It is important that the user to make this decision instead of automatic, implicit conversion
 - channels not deserialized correctly
 - channelIndex not deserialized correctly
 - converter not deserialized correctly
 - checked for wrong source class
The lookup is always provided in the Paintera dataset backend, no need to deserialize when converting.
Currently, there is still concurrent exceptions. Fix those, then get this ready for discussion!
Still experiencing concurrent modification exceptions
Before adding the data to the table, it is sorted already at the moment
@hanslovsky
Copy link
Collaborator Author

hanslovsky commented Jan 10, 2020

Unfortunately, I do not think, that this will work without an explicit feature store concept. Trying to extract the counts for a larger block than CREMI (e.g. lauritzen 03) takes forever and freezes on my relatively powerful laptop. In order to proceed with this, we will need to create a fragment feature store that is pre-computed (e.g. when converting into Paintera data format) with only targeted updates after painting.

This feature store can be either (1) a database or (2) N5 based

Navigate to next smallest segment that intersects with at least one synapse.

Does not work as of now, will probably require a more well-thoughtout feature store.
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.

1 participant