One mode ellipsoid data

[jadie1]

Change ellipsoid data to only vary along x axis. Working on the one_mode_ellipsoid branch.

• Update shape cohort generation code so user can specify which axis should vary
• Create ellipsoid_1mode data and put on portal with readme (use-case-data-v2/ellipsoid_1mode/)
• Check ellisoid_1mode into github and remove ellisoid-v0 and ellipsoid_fd-v0
• Create ellipsoid_1mode_aligned data and put on portal with readme (use-case-data-v2/ellipsoid_1mode_aligned/)
• Update all use cases
  • Update ellipsoid use case to use ellipsoid_1mode segs
  • Update ellipsoid_cut use case to use ellipsoid_1mode_aligned segs (issue ellipsoid_cut model shows poor correspondence #1104)
  • Update ellipsoid_mesh use case to use ellipsoid_1mode_aligned meshes
  • Update ellipsoid_fd to use ellipsoid_1mode segs
  • Verify ellipsoid_evaluate use case still works as expected
• Update portal data to have all use case output (all but ellipsoid_cut added issue ellipsoid_cut model shows poor correspondence #1104)
• Update all notebooks
  • Update getting-started-with-segmentations.ipynb to use ellipsoid_1mode data
  • Update getting-started-with-meshes.ipynb to use ellipsoid_1mode data
  • Update getting-started-with-exploring-segmentations.ipynb to use ellipsoid_1mode data
  • Update getting-started-with-grooming-segmentations.ipynb to use ellipsoid_1mode data
  • Update getting-started-with-shape-cohort-generation.ipynb for changes to shape cohort generation
• Update documentation for getting-started-with-shape-cohort-generation.ipynb

[jadie1]

@sheryjoe Could you review this and let me know if I missed anything?

[jadie1]

@akenmorris Since none of the use cases will be using ellipsoid-v0 now I will remove it from github. Should I add the new dataset to github or should we require the users download it from the portal?

[akenmorris]

If it's small enough, it would be great to include it. It's nice to be able to run a use case without logging in. It will just go to LFS.

[jadie1]

@sheryjoe Do you think we need to replace every video and png in the notebooks? If the output looks similar but a little different is that okay?

[jadie1]

@sheryjoe Here is an example of the output using the new data compared to the video:

Do you think it is okay if the output is similar but not identical? If not we need to start re-recording the ~40 mp4s.

[sheryjoe]

@jadie1 given similar output, i don't think we have to update the videos.

[sheryjoe]

@jadie1 please keep in mind that the ellispoid_fd and ellispoid_cut use cases assume aligned ellipsoids (as far as I recall). They were using ellispoid-v0

[jadie1]

@jadie1 please keep in mind that the ellispoid_fd and ellispoid_cut use cases assume aligned ellipsoids (as far as I recall). They were using ellispoid-v0

Yes they are now using the ellispoid_1mode_aligned dataset which is essentially the same as ellipsiod_v0, just generated using the shape cohort generator package.

[jadie1]

Here are the results using the new data.

ellipsoid

ellipsoid_fd

ellipsoid_mesh

ellipsoid_cut

(Issue #1104)

[akenmorris]

What branch is this on?

[iyerkrithika21]

What branch is this on?

one_mode_ellipsoid

[akenmorris]

Ok, the only thing remaining is the ellipsoid_cut params?

[iyerkrithika21]

Yes
#1073 (comment)