Tensorflow segmentation statistics #3946
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
daniel-wer
pushed a commit
that referenced
this pull request
Apr 10, 2019
* integrate first version of tensorflow-js segmentation model * add flood fill and whole viewport prediction * implement high-perf way of accessing arbitrary cuboid data in front-end * fix some things, wand is working again * fix everything * copy tf models via webpack * upgrade webpack * increase node memory * fix lint, tests * fix model asset path * update tensorflowjs to 1.0 * move viewport when pressing alt and moving mouse * Tensorflow segmentation statistics (#3946) * [WIP] provide sampled mean and stdev for color layer data * convert to unsigned, clean up * best resolution is actually min, not max * pretty-backend * adapt tensorflow segmentation to rectangular viewports * fix lint * added mean and stdDev fetching route and using it for the magic wand tool * applied new formular again * fixed lint and flow * predict for next 5 slices as well * allow tf inference in webworker * made magic wand tool optional * fix shard in webpack-dev-server mode * enabled magic wand tool as default * fix tests * use new segem model, restrict viewport to 100x100 voxel for perf * small perf/cleanup fixes * load model from right path * only use webworker if offscreencanvas is supported * use 3d-floodfill for magic wand * add and adapt floodfill lib, do 3d floodfill, do tile-based predict and floodfill * adapt floodfill to work slice for slice, optimize voxel labeling, remove unneccessary floodfill parts * add flow to floodfill, fix flow * add magic wand unlimited mode, will predict next x slices until aborted via toast * use global tiles instead of viewport local, makes everything easier * update the whole position, so that segments that move out of the viewport later will be correctly predicted * fix magic wand for xz and yz planes * fix mean,stddev memoization, refactor common code * small changes to toast and code * add new tensorflow model, apply PR feedback #1 * apply PR feedback #2 * update redux-saga and typedefs to 1.0, avoid spawning magic wand twice * clean up prediction data for old z slices, rename magic wand to automatic brush * refactor automatic brushing code into own module * disable magic wand by default * react to zoom changes, amend automatic brush label * apply PR feedback #3, use newest model with slightly different normalization, make floodfill_threshold configurable * disable automatic brush by default
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.
URL of deployed dev instance (used for testing):
Steps to test:
http://localhost:9000/data/datasets/Connectomics_Department/Sample_e2006_wkw/layers/color/colorStatistics?token=secretScmBoyTokenshould contain two json fieldsmeanandstdDev