v0.0.3

New changes include:

• Increase ransac iterations to 8000
• Added an additional case for fixed y vector prediction (this emerges because R(x \times y) \neq Rx \times Ry
• Estimation of accuracy is more robust (now, only takes keypoint ids and doesn't care about which rows they appear in)

v0.0.2

• Add three subpanels for detecting nuclei, estimating transform and evaluating metrics
• Next todo's include:
  • Add haussdarff distance metric
  • Add tick and refresh buttons and icons for updating progress of actions on pushbuttons

v0.0.1

• This release includes thejava code originally used for evaluating results at the time of the BIC workshop submission
• Additionally, it includes the equivalent python code used and the first attempts of deploying it in napari
• The entire Platynereis dataset for which results were reported in the publication is included as release assets here.
  • Directory 01-insitus contains confocal images of 12 independent specimens (labeled as 02, 04, 05, 06, 07, 08, 10, 11, 12, 13, 14 and 16). Additionally nuclei centroids and locations of 12 keypoints are provided for each specimen
  • Directory 02-live contains 12 static snapshots of a live embryo (again labeled as 02, 04, 05, 06, 07, 08, 10, 11, 12, 13, 14 and 16), imaged under light sheet microscopy. Additionally nuclei centroids and locations of 12 keypoints are provided for each snapshot
  • Folder with the same name in the two modalities have equivalent development age and should be registered with each other
  • All detections and keypoints are available in isotropic global coordinates as csv files are are specified in the column order xyz
• In v0.0.2, the python code gets further refactored