Add PseudoLabeling transform for unlabeled dataset#1594
Add PseudoLabeling transform for unlabeled dataset#1594sooahleex merged 4 commits intoopen-edge-platform:developfrom
Conversation
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## develop #1594 +/- ##
===========================================
+ Coverage 81.06% 81.07% +0.01%
===========================================
Files 278 277 -1
Lines 32517 32592 +75
Branches 6607 6623 +16
===========================================
+ Hits 26360 26425 +65
- Misses 4701 4711 +10
Partials 1456 1456
Flags with carried forward coverage won't be shown. Click here to find out more. ☔ View full report in Codecov by Sentry. |
faff21b to
2e5415f
Compare
2e5415f to
2d796d1
Compare
7ab7d2f to
ad83fa3
Compare
c03546d to
19a7d16
Compare
|
|
||
| pseudo = np.array(self._labels)[ind][0] | ||
| pseudo_annotation = [Label(label=self._label_indices[pseudo])] | ||
| return self.wrap_item(item, annotations=pseudo_annotation) |
There was a problem hiding this comment.
It seems that the original annotations are replaced with the pseudo annotation. Is it intended? Then how about add this notice to the document?
There was a problem hiding this comment.
In this case, I assumed the case that each item does not have any annotation. so the original annotations is not existed. If my assumption is not proper, please let me know.
There was a problem hiding this comment.
Oh, I see. Then, we can enhance this feature when the assumption is changed.
Summary
PseudoLabelingtransform for unlabeled datasetExplorerof Datumaro to make pseudo label for each datasetitemHow to test
Checklist
License
Feel free to contact the maintainers if that's a concern.