fix: remove black stripe artefacts when axis != -1 in
The clean-up loop that sets un-normalised channels to 0 expected the channel dimension to be last. However, since np.moveaxis(..., -1, axis) was done prior to this, instead of setting the channel to zero, it would set a row/column (depending on the axis) to zero instead, resulting in a black bar across the image.
Looks good. What were the settings/image dimensions that produced the error?
It happens when you pass in paths to the evaluation method. It loads the images, makes them channels first, and requests the normalisation function with axis=0. It happens whenever axis!=-1 I think.
Codecov Report
All modified and coverable lines are covered by tests :white_check_mark:
Project coverage is 48.89%. Comparing base (
8483650) to head (9649909). Report is 12 commits behind head on main.
Additional details and impacted files
@@ Coverage Diff @@
## main #1270 +/- ##
==========================================
- Coverage 48.96% 48.89% -0.07%
==========================================
Files 16 16
Lines 3762 3767 +5
==========================================
Hits 1842 1842
- Misses 1920 1925 +5
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
:rocket: New features to boost your workflow:
- :snowflake: Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
Will this be merged? Not trying to rush, just curious about the release schedule