You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently Auto Annotation's run_model.py has a way to visualize the results, but it doesn't have any way to visualize labels. This has lead to off-by-1 errors in development.
Note that when using the --serialize command, the text label is actually overwritten (for reasons that are explained in the note starting at this link) in the following line:
The above overriden data would need to be preserved by some programmatic means to preserve the plain english descriptions. A simple dictionary copy before re-writing at line 126 would likely suffice.
The text was updated successfully, but these errors were encountered:
Currently Auto Annotation's
run_model.py
has a way to visualize the results, but it doesn't have any way to visualize labels. This has lead to off-by-1 errors in development.Ref: #912
Ref: #890 (comment)
Need to add the ability to visualize labels to ensure the mapping.json remains accurate.
This should be relatively easy. See below for a reference implementation:
https://github.com/opencv/open_model_zoo/blob/e372d4173e50741a6828cda415d55c37320f89cd/demos/python_demos/object_detection_demo_yolov3_async/object_detection_demo_yolov3_async.py#L318
Note that when using the
--serialize
command, the text label is actually overwritten (for reasons that are explained in the note starting at this link) in the following line:https://github.com/opencv/cvat/blob/32027ce884c0584015874c4ae99ba7f53ffb46c0/utils/auto_annotation/run_model.py#L126
The above overriden data would need to be preserved by some programmatic means to preserve the plain english descriptions. A simple dictionary copy before re-writing at line 126 would likely suffice.
The text was updated successfully, but these errors were encountered: