Skip to content

146 birdnet to labelstudio preannotated flow - #149

Open
kgarwoodsdzwa wants to merge 11 commits into
devfrom
146_birdnet_to_labelstudio_preannotated_flow
Open

146 birdnet to labelstudio preannotated flow#149
kgarwoodsdzwa wants to merge 11 commits into
devfrom
146_birdnet_to_labelstudio_preannotated_flow

Conversation

@kgarwoodsdzwa

Copy link
Copy Markdown
Member

No description provided.

Specific to the example run. Pkl is no longer needed, can be
csv. And the replacement of the filepath is specific but
should be a feature because it's a common issue
I could see us making a similar function that doesn't output the dict
so I felt it made sense to specify. The dict is needed to upload the
predictions to labelstudio. However, you don't always do that. In that
case, you might want some other metadata in some other form that isnt
that specifically formatted dict.
Later when we want to generate samples from the labelstudio
labels, we want to be able to trace the segment back to the original
clip, and the original offset and duration of the clip from
the original file. this way when we generate clips to use for training
we have the additional context of the original file in case we need it
when window expanding
Labelstudio will have a different basepath
in the reslts data when we export the labels so we don't need
the basepath for the segment when we create it locally.
That can be something done later but not needed for this pr
When using this metadata elsewhere, we need the column names to match
I changed it downstream because the column name made more sense, but
it needed to be updated here as well because they need to match
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant