Import multiple DLC datasets from a folder #437
Conversation
Codecov Report
@@ Coverage Diff @@
## develop #437 +/- ##
========================================
Coverage 60.40% 60.41%
========================================
Files 119 119
Lines 17945 17979 +34
========================================
+ Hits 10840 10862 +22
- Misses 7105 7117 +12
Continue to review full report at Codecov.
|
sleap/gui/commands.py
Outdated
| print("Merged labels:") | ||
| merged_labels.describe() | ||
|
|
||
| merged_labels.save("{}.slp".format(params["filename"])) | ||
| merged_labels.save("{}.pkg.slp".format(params["filename"]), | ||
| with_images=True) # comment this out to skip saving with embedded images |
There was a problem hiding this comment.
Remove this -- the labels will be loaded from memory
sleap/gui/commands.py
Outdated
| @staticmethod | ||
| def do_action(context: "CommandContext", params: dict): | ||
|
|
||
| csv_files = glob("{}/*/*.csv".format(params["dirname"])) |
There was a problem hiding this comment.
Nit: Prefer f-strings to .format(), i.e., glob(f"{params['dirname']}/*/*.csv")
sleap/gui/commands.py
Outdated
| if len(dirname) == 0: | ||
| return False | ||
| params["dirname"] = dirname | ||
| params["filename"] = "converted_dlc_dataset" |
There was a problem hiding this comment.
See above -- filename not needed. This op should not do any I/O. User can save the labels from File -> Save after importing
| coords,x,y,x,y,x,y,x,y,x,y,x,y | ||
| labeled-data/video/img000.png,0,1,2,3,4,5,6,7,8,9,10,11 | ||
| labeled-data/video/img001.png,12,13,,,15,16,17,18,,,20,21 | ||
| labeled-data/video/img002.png,22,23,24,25,26,27,,,,,, |
There was a problem hiding this comment.
Maybe change a couple of these numbers so we can test that we're not just importing the same thing twice?
There was a problem hiding this comment.
Split the 2 files to have a bit more realistic test. One has labels from the first 2 frames, the other one from the last frame. Also added the images for manual verification via UI
|
PTAL |
Import multiple DLC datasets in a folder.
Assuming a folder structure like this:
user can select the
'labeled-data'folder in the dialog, and then all'*.csv'files 2 levels below will be imported and merged into a singleSleapdataset.Fixes #412