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

feat: replace forward pass with decision tree #8

Merged
merged 27 commits into from
May 25, 2019

Conversation

william-silversmith
Copy link
Contributor

@william-silversmith william-silversmith commented May 22, 2019

Elaboration into 3D of the Wu et al 2005 paper

Wu, Otoo, and Suzuki. "Two Strategies to Speed up
Connected Component Labeling Algorithms". 2005. LBNL-59102.

https://crd-legacy.lbl.gov/~kewu/ps/LBNL-59102.html

Resolves #7 and Resolves #9

Elaboration into 3D of the Wu et al 2005 paper

Wu, Otoo, and Suzuki. "Two Strategies to Speed up
Connected Component Labeling Algorithms". 2005. LBNL-59102.

https://crd-legacy.lbl.gov/~kewu/ps/LBNL-59102.html
@william-silversmith william-silversmith added the feature New feature or request label May 22, 2019
@william-silversmith william-silversmith self-assigned this May 22, 2019
@william-silversmith william-silversmith added the bug Something isn't working label May 23, 2019
@william-silversmith
Copy link
Contributor Author

Still a random segfault hiding in there somewhere... Also, eyeballing the output it looks pretty good, but the number of segments is different compared to master.

@william-silversmith
Copy link
Contributor Author

In time trials, so far it looks like on a particular 512 cube, the old code takes about 7 to 8 seconds while this version takes about 3 to 4 seconds. Not bad, provided I can wring the last few issues out of it.

@william-silversmith william-silversmith changed the title wip: replace forward pass with decision tree feat: replace forward pass with decision tree May 24, 2019
@william-silversmith
Copy link
Contributor Author

Realized that I had been thinking about the problem wrong. 1) The mask needs to essentially cover half the volume. 2) The Z = 0 part is just Kesheng Wu's algorithm, the 3D part is handling Z = -1. This is a very elegant decomposition of the problem.

Initial tests of the new decision tree look promising, and somehow it was even faster on the test volume - 2.8 seconds!

@william-silversmith william-silversmith merged commit 0980366 into master May 25, 2019
@william-silversmith william-silversmith deleted the wms_decision_tree branch May 25, 2019 01:45
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
bug Something isn't working feature New feature or request
Projects
None yet
Development

Successfully merging this pull request may close these issues.

Not actually 26 connected? Better Algorithm
1 participant